feat: initial release of HyperSynapse-SAS multi-model & universal chat templates

README.md

@@ -0,0 +1,157 @@
+---
+license: mit
+language:
+- en
+- zh
+tags:
+- chat-template
+- jinja
+- hypersynapse-sas
+- cognitive-architecture
+- reasoning
+- multi-agent
+- swarm
+- deepseek
+- qwen
+- glm
+- llama
+- mistral
+- gemma
+- universal
+pipeline_tag: text-generation
+---
+
+<div align="center">
+
+# 🧠 HyperSynapse-SAS (SelfAgentSwarm) Chat Templates
+### *The Unified 10-Level Cognitive Escalation & Autonomous Multi-Agent Swarm Framework*
+
+[![Solstice AI](https://img.shields.io/badge/Solstice--AI-HyperSynapse--SAS-blueviolet?style=for-the-badge&logo=openai)](https://huggingface.co/Solstice-AI)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg?style=for-the-badge)](https://opensource.org/licenses/MIT)
+[![Model-Agnostic](https://img.shields.io/badge/Architecture-Model--Agnostic-success?style=for-the-badge)](https://huggingface.co/Solstice-AI/HyperSynapse-SAS-Chat-Templates)
+[![Levels](https://img.shields.io/badge/Cognitive%20Levels-0%20to%209%20Solo-orange?style=for-the-badge)](https://huggingface.co/Solstice-AI/HyperSynapse-SAS-Chat-Templates)
+[![Swarm](https://img.shields.io/badge/Swarm-20--Agent%20%2B%20Council-red?style=for-the-badge)](https://huggingface.co/Solstice-AI/HyperSynapse-SAS-Chat-Templates)
+
+</div>
+
+---
+
+## 🌟 Overview
+
+**HyperSynapse-SAS** (HyperSynapse Self-Agent Swarm) is a high-performance cognitive chat template framework that equips any Large Language Model with:
+1. **10-Level Solo Cognitive Escalation (`Level 0 Mortal` to `Level 9 Oracle`)**: Dynamically scale reasoning depth, token runway, and proof rigor per prompt without triggering multi-persona roleplay clutter.
+2. **Decoupled Multi-Agent Swarm Modes (`[swarm]` & `[deep-swarm]`)**: On-demand simulated 20-agent divergent swarms and adversarial consensus councils for massive, open-ended problem spaces.
+3. **Model-Agnostic / Universal Template**: A plug-and-play ChatML Jinja template compatible with **any modern LLM**, plus dedicated, fine-tuned templates for all major model families.
+
+---
+
+## 📁 Available Templates
+
+| Template | File | Target Architectures & Families |
+| :--- | :--- | :--- |
+| **Universal (Model-Agnostic)** | [`templates/universal.jinja`](templates/universal.jinja) | Standard ChatML, vLLM, SGLang, llama.cpp, Ollama, LM Studio, Any LLM |
+| **DeepSeek** | [`templates/deepseek.jinja`](templates/deepseek.jinja) | DeepSeek-V3, DeepSeek-V4, DeepSeek-V4.1 Flash, DSML Tool Calling |
+| **Qwen** | [`templates/qwen.jinja`](templates/qwen.jinja) | Qwen 2.5, Qwen 3, Qwen 3.5, Qwen 3.8, Qwopus, QwQ |
+| **GLM** | [`templates/glm.jinja`](templates/glm.jinja) | GLM-4, GLM-5.3, GLM Flash |
+| **Llama** | [`templates/llama3.jinja`](templates/llama3.jinja) | Meta Llama 3, Llama 3.1, Llama 3.2, Llama 3.3, Llama 4 |
+| **Mistral** | [`templates/mistral.jinja`](templates/mistral.jinja) | Mistral 7B/12B, Mixtral 8x7B/8x22B, Codestral, Pixtral |
+| **Gemma** | [`templates/gemma.jinja`](templates/gemma.jinja) | Google Gemma 2, Gemma 3, Gemma 4 |
+
+---
+
+## 🎯 Cognitive Trigger Spectrum
+
+Simply prepend a tag to your query or pass `--level X` in your system instructions:
+
+### Solo Cognitive Escalation (Pure Analytical Focus)
+
+| Tag / Level | Persona / Archetype | Budget | Cognitive Modality |
+| :--- | :--- | :--- | :--- |
+| **`[Level 0]`** | **Mortal** | 0 tokens | **Instant Response**: CoT/thinking completely disabled. Rapid, direct execution with zero discursive fluff. |
+| **`[Level 1]`** | **Hermes** | ~256 tokens | **Rapid Instinct**: Single-path heuristic sanity check before emitting answer. |
+| **`[Level 2]`** | **Apollo** | ~512 tokens | **Crisp Logic**: Explicit step-by-step reasoning with edge-case validation. |
+| **`[Level 3]`** | **Artemis** | ~1,024 tokens | **Boundary Hunter**: Focuses on edge cases, off-by-one errors, and null states. |
+| **`[Level 4]`** | **Athena** | ~2,048 tokens | **Systemic Balance**: Second-order effects, system architecture, modular design. |
+| **`[Level 5]`** | **Prometheus** | ~4,096 tokens | **Defensive Engineering**: Fault tolerance, production bottlenecks, failure mode audit. |
+| **`[Level 6]`** | **Solstice** | ~8,192 tokens | **Deep Derivation**: Exhaustive theoretical derivation and mathematical mechanics. |
+| **`[Level 7]`** | **Hyperion (Default)** | ~16,384 tokens | **Foundational Synthesis**: First-principles breakdown from ground axioms. |
+| **`[Level 8]`** | **Einstein** | ~32,768 tokens | **Mastery & Invariants**: Theoretical proofs, deep algorithmic complexity analysis. |
+| **`[Level 9]`** | **Oracle** | Unbounded | **Maximal Rigor**: Exhaustive search runway, proof-grade verification, definitive closure. |
+
+### Autonomous Multi-Agent Swarm Modes
+
+| Trigger | Mode Name | Protocol & Behavior |
+| :--- | :--- | :--- |
+| **`[swarm]`** | **20-Agent Divergent Swarm** | Activates a simulated 20-perspective emergent swarm (Analytical Logician, Creative Maverick, Systems Engineer, Red-Teamer, Optimizer, Critic, Synthesizer) with divergent exploration followed by structured synthesis. |
+| **`[deep-swarm]`** | **Adversarial Research Council** | Deploys a multi-round debate tournament pitting 3–5 specialized domain specialists against each other to falsify weak assumptions and converge on proof-grade deliverables. |
+
+---
+
+## 🚀 Quick Start
+
+### 1. Using the Universal Template in Python (`transformers`)
+```python
+from transformers import AutoTokenizer
+
+tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-7B-Instruct")
+
+# Load the universal HyperSynapse-SAS template
+with open("templates/universal.jinja") as f:
+    tokenizer.chat_template = f.read()
+
+messages = [
+    {"role": "user", "content": "[Level 3] Explain quantum decoherence rigorously."}
+]
+
+prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+print(prompt)
+```
+
+### 2. Apply Directly to Any Hugging Face Hub Repo
+Use our built-in CLI applier to update any repository or local folder with one command:
+```bash
+# Push to Hugging Face Model Hub
+python apply_template.py --template deepseek --model-id Solstice-AI/My-Model --push
+
+# Or apply locally to a weights directory
+python apply_template.py --template universal --local-dir ./my_local_model
+```
+
+### 3. Using in `vLLM` or `SGLang`
+Pass `--chat-template templates/universal.jinja` directly at startup:
+```bash
+vllm serve meta-llama/Llama-3.1-8B-Instruct \
+  --chat-template templates/universal.jinja \
+  --port 8000
+```
+
+### 4. Using in `llama.cpp`
+```bash
+llama-cli \
+  --model my_model.gguf \
+  --chat-template-file templates/universal.jinja \
+  -p "<|im_start|>user\n[swarm] Architect an ultra-low latency event bus.<|im_end|>\n<|im_start|>assistant\n"
+```
+
+### 5. Ollama Modelfile
+```dockerfile
+FROM ./my_model.gguf
+TEMPLATE """{{ if .System }}<|im_start|>system
+{{ .System }}<|im_end|>
+{{ end }}{{ range .Messages }}<|im_start|>{{ .Role }}
+{{ .Content }}<|im_end|>
+{{ end }}<|im_start|>assistant
+"""
+PARAMETER stop "<|im_end|>"
+```
+
+---
+
+## 📄 License
+
+Released under the permissive **MIT License**. Free for commercial and personal use across any open-weight model or inference stack.
+
+<div align="center">
+<b>Developed by Solstice-AI</b><br>
+<i>Unshackled Cognition • Multi-Agent Collective Intelligence • Scalable Autonomous Systems</i>
+</div>

apply_template.py

@@ -0,0 +1,103 @@
+#!/usr/bin/env python3
+"""
+HyperSynapse-SAS Chat Template Applier
+Usage:
+    python apply_template.py --template universal --local-dir ./path/to/model
+    python apply_template.py --template deepseek --model-id Solstice-AI/my-model --push
+"""
+
+import argparse
+import json
+import os
+import sys
+from huggingface_hub import HfApi, hf_hub_download
+
+TEMPLATES = {
+    "universal": "templates/universal.jinja",
+    "deepseek": "templates/deepseek.jinja",
+    "qwen": "templates/qwen.jinja",
+    "glm": "templates/glm.jinja",
+    "llama3": "templates/llama3.jinja",
+    "llama": "templates/llama3.jinja",
+    "mistral": "templates/mistral.jinja",
+    "gemma": "templates/gemma.jinja",
+}
+
+def load_template(name):
+    base_dir = os.path.dirname(os.path.abspath(__file__))
+    rel_path = TEMPLATES.get(name.lower())
+    if not rel_path:
+        raise ValueError(f"Unknown template: {name}. Available: {list(TEMPLATES.keys())}")
+    full_path = os.path.join(base_dir, rel_path)
+    with open(full_path, "r", encoding="utf-8") as f:
+        return f.read()
+
+def apply_local(local_dir, template_name):
+    tc_path = os.path.join(local_dir, "tokenizer_config.json")
+    if not os.path.exists(tc_path):
+        print(f"Error: {tc_path} not found.")
+        sys.exit(1)
+    
+    template_code = load_template(template_name)
+    with open(tc_path, "r", encoding="utf-8") as f:
+        tc = json.load(f)
+    
+    tc["chat_template"] = template_code
+    with open(tc_path, "w", encoding="utf-8") as f:
+        json.dump(tc, f, indent=2, ensure_ascii=False)
+        f.write("\n")
+    
+    # Also save chat_template.jinja standalone
+    jinja_path = os.path.join(local_dir, "chat_template.jinja")
+    with open(jinja_path, "w", encoding="utf-8") as f:
+        f.write(template_code)
+    
+    print(f" Successfully applied {template_name} template to {local_dir}")
+
+def apply_hub(model_id, template_name, push=False, token=None):
+    api = HfApi(token=token)
+    print(f"Downloading tokenizer_config.json from {model_id}...")
+    tc_path = hf_hub_download(repo_id=model_id, filename="tokenizer_config.json", token=token)
+    with open(tc_path, "r", encoding="utf-8") as f:
+        tc = json.load(f)
+    
+    template_code = load_template(template_name)
+    tc["chat_template"] = template_code
+    
+    if push:
+        print(f"Pushing updated tokenizer_config.json and chat_template.jinja to {model_id}...")
+        api.upload_file(
+            path_or_fileobj=json.dumps(tc, indent=2, ensure_ascii=False).encode("utf-8"),
+            path_in_repo="tokenizer_config.json",
+            repo_id=model_id,
+            commit_message=f"feat: apply HyperSynapse-SAS {template_name} chat template"
+        )
+        api.upload_file(
+            path_or_fileobj=template_code.encode("utf-8"),
+            path_in_repo="chat_template.jinja",
+            repo_id=model_id,
+            commit_message=f"feat: add HyperSynapse-SAS {template_name} chat_template.jinja"
+        )
+        print(f" Successfully pushed to {model_id}!")
+    else:
+        print("Dry run completed. Pass --push to commit changes to Hugging Face Hub.")
+
+def main():
+    parser = argparse.ArgumentParser(description="Apply HyperSynapse-SAS Chat Templates")
+    parser.add_argument("--template", "-t", required=True, choices=list(TEMPLATES.keys()), help="Template family to apply")
+    parser.add_argument("--local-dir", "-l", help="Path to local model directory")
+    parser.add_argument("--model-id", "-m", help="Hugging Face Model ID (e.g. org/model)")
+    parser.add_argument("--push", action="store_true", help="Push to Hugging Face Hub")
+    parser.add_argument("--token", help="Hugging Face User Access Token")
+    
+    args = parser.parse_args()
+    if args.local_dir:
+        apply_local(args.local_dir, args.template)
+    elif args.model_id:
+        apply_hub(args.model_id, args.template, push=args.push, token=args.token)
+    else:
+        print("Please provide either --local-dir or --model-id.")
+        sys.exit(1)
+
+if __name__ == "__main__":
+    main()

generate_family_templates.py

@@ -0,0 +1,106 @@
+import re
+import jinja2
+
+# 1. Generate universal.jinja from qwen.jinja base with universal tags
+with open("/Users/shreyangondaliya/solstice_quants/HyperSynapse-SAS-Chat-Templates/templates/qwen.jinja") as f:
+    qwen_content = f.read()
+
+# For universal.jinja:
+# Header:
+universal_header = """{#- ================================================================= -#}
+{#- HyperSynapse-SAS (SelfAgentSwarm) 10-Level Cognitive Architecture -#}
+{#- Model-Agnostic / Universal ChatML Chat Template                   -#}
+{#- Compatible with any LLM runtime (llama.cpp, vLLM, SGLang, Ollama) -#}
+{#- Levels: 0=Mortal/Disabled, 1=Hermes/ulow, 2=Apollo/low,           -#}
+{#- 3=Artemis/lmed, 4=Athena/medium, 5=Prometheus/mhigh,              -#}
+{#- 6=Solstice/high, 7=Hyperion/xhigh (Default), 8=Einstein (Solo),   -#}
+{#- 9=Oracle (Solo). Swarm: swarm (20-Agent), deep-swarm (Council)    -#}
+{#- ================================================================= -#}
+"""
+
+universal_content = universal_header + qwen_content[qwen_content.find("{%- set inst_hermes -%}"):]
+with open("/Users/shreyangondaliya/solstice_quants/HyperSynapse-SAS-Chat-Templates/templates/universal.jinja", "w") as f:
+    f.write(universal_content)
+
+print("Created universal.jinja!")
+
+# 2. Build Llama 3 / 4 template
+# In Llama 3 / 4, special tokens are:
+# <|begin_of_text|>
+# <|start_header_id|>system<|end_header_id|>\n\n...<|eot_id|>
+# <|start_header_id|>user<|end_header_id|>\n\n...<|eot_id|>
+# <|start_header_id|>assistant<|end_header_id|>\n\n...<|eot_id|>
+
+llama_header = """{#- ================================================================= -#}
+{#- HyperSynapse-SAS (SelfAgentSwarm) 10-Level Cognitive Architecture -#}
+{#- Tailored for Meta Llama 3 / 3.1 / 3.2 / 3.3 / Llama 4             -#}
+{#- Levels: 0=Mortal/Disabled, 1=Hermes/ulow, 2=Apollo/low,           -#}
+{#- 3=Artemis/lmed, 4=Athena/medium, 5=Prometheus/mhigh,              -#}
+{#- 6=Solstice/high, 7=Hyperion/xhigh (Default), 8=Einstein (Solo),   -#}
+{#- 9=Oracle (Solo). Swarm: swarm (20-Agent), deep-swarm (Council)    -#}
+{#- ================================================================= -#}
+"""
+
+# Replace ChatML tokens in qwen rendering section with Llama tokens
+qwen_instructions_part = qwen_content[qwen_content.find("{%- set inst_hermes -%}"):qwen_content.find("{%- set ns = namespace(")]
+qwen_logic_part = qwen_content[qwen_content.find("{%- set ns = namespace("):]
+
+# Adapt rendering logic for Llama
+llama_render = qwen_logic_part.replace("<|im_start|>system\\n", "<|start_header_id|>system<|end_header_id|>\\n\\n")
+llama_render = llama_render.replace("<|im_start|>user\\n", "<|start_header_id|>user<|end_header_id|>\\n\\n")
+llama_render = llama_render.replace("<|im_start|>assistant\\n", "<|start_header_id|>assistant<|end_header_id|>\\n\\n")
+llama_render = llama_render.replace("<|im_start|>developer\\n", "<|start_header_id|>developer<|end_header_id|>\\n\\n")
+llama_render = llama_render.replace("<|im_end|>\\n", "<|eot_id|>\\n")
+llama_render = llama_render.replace("<|im_end|>", "<|eot_id|>")
+
+llama_content = llama_header + qwen_instructions_part + "{{- bos_token -}}\n" + llama_render
+with open("/Users/shreyangondaliya/solstice_quants/HyperSynapse-SAS-Chat-Templates/templates/llama3.jinja", "w") as f:
+    f.write(llama_content)
+
+print("Created llama3.jinja!")
+
+# 3. Build Mistral template
+mistral_header = """{#- ================================================================= -#}
+{#- HyperSynapse-SAS (SelfAgentSwarm) 10-Level Cognitive Architecture -#}
+{#- Tailored for Mistral / Mixtral / Codestral / Pixtral              -#}
+{#- Levels: 0=Mortal/Disabled, 1=Hermes/ulow, 2=Apollo/low,           -#}
+{#- 3=Artemis/lmed, 4=Athena/medium, 5=Prometheus/mhigh,              -#}
+{#- 6=Solstice/high, 7=Hyperion/xhigh (Default), 8=Einstein (Solo),   -#}
+{#- 9=Oracle (Solo). Swarm: swarm (20-Agent), deep-swarm (Council)    -#}
+{#- ================================================================= -#}
+"""
+# Mistral uses [INST] and [/INST]
+mistral_render = qwen_logic_part.replace("<|im_start|>system\\n", "[INST] ")
+mistral_render = mistral_render.replace("<|im_start|>user\\n", "[INST] ")
+mistral_render = mistral_render.replace("<|im_start|>assistant\\n", "")
+mistral_render = mistral_render.replace("<|im_end|>\\n", " [/INST]\n")
+mistral_render = mistral_render.replace("<|im_end|>", " [/INST]")
+
+mistral_content = mistral_header + qwen_instructions_part + "{{- bos_token -}}\n" + mistral_render
+with open("/Users/shreyangondaliya/solstice_quants/HyperSynapse-SAS-Chat-Templates/templates/mistral.jinja", "w") as f:
+    f.write(mistral_content)
+
+print("Created mistral.jinja!")
+
+# 4. Build Gemma template
+gemma_header = """{#- ================================================================= -#}
+{#- HyperSynapse-SAS (SelfAgentSwarm) 10-Level Cognitive Architecture -#}
+{#- Tailored for Google Gemma 2 / Gemma 3 / Gemma 4                   -#}
+{#- Levels: 0=Mortal/Disabled, 1=Hermes/ulow, 2=Apollo/low,           -#}
+{#- 3=Artemis/lmed, 4=Athena/medium, 5=Prometheus/mhigh,              -#}
+{#- 6=Solstice/high, 7=Hyperion/xhigh (Default), 8=Einstein (Solo),   -#}
+{#- 9=Oracle (Solo). Swarm: swarm (20-Agent), deep-swarm (Council)    -#}
+{#- ================================================================= -#}
+"""
+gemma_render = qwen_logic_part.replace("<|im_start|>system\\n", "<start_of_turn>system\\n")
+gemma_render = gemma_render.replace("<|im_start|>user\\n", "<start_of_turn>user\\n")
+gemma_render = gemma_render.replace("<|im_start|>assistant\\n", "<start_of_turn>model\\n")
+gemma_render = gemma_render.replace("<|im_start|>developer\\n", "<start_of_turn>system\\n")
+gemma_render = gemma_render.replace("<|im_end|>\\n", "<end_of_turn>\\n")
+gemma_render = gemma_render.replace("<|im_end|>", "<end_of_turn>")
+
+gemma_content = gemma_header + qwen_instructions_part + "{{- bos_token -}}\n" + gemma_render
+with open("/Users/shreyangondaliya/solstice_quants/HyperSynapse-SAS-Chat-Templates/templates/gemma.jinja", "w") as f:
+    f.write(gemma_content)
+
+print("Created gemma.jinja!")

templates/deepseek.jinja

@@ -0,0 +1,304 @@
+{#- ================================================================= -#}
+{#- Solstice-AI 10-Level Cognitive Reasoning & Instruct Chat Template -#}
+{#- Tailored for DeepSeek-V4.1 Native Tokenizer & DSML Tool Calling  -#}
+{#- Core Spectrum: 0=Mortal, 1=Hermes, 2=Apollo, 3=Artemis, 4=Athena -#}
+{#- 5=Prometheus, 6=Solstice, 7=Hyperion (Default), 8=Einstein,       -#}
+{#- 9=Oracle. Separate Swarms: swarm (20-Agent), deep-swarm (Council) -#}
+{#- ================================================================= -#}
+{%- set inst_hermes -%}
+Reasoning Effort: Rapid Instinct (Level 1 — Hermes).
+Role & Persona: You are Hermes, the swift cognitive vanguard. Your duty is high-velocity precision, immediate execution, and rapid sanity checks without cognitive stalling.
+Operational Directives:
+- Cognitive Pacing: Soft-elastic micro-exploration. For direct inquiries, conclude your thinking rapidly. If subtle contradictions emerge, resolve them with minimal overhead.
+- Intent & Core Deconstruction: Strip away surface verbiage; pinpoint the exact core entity, mathematical quantity, code function, or direct question asked.
+- Direct Route to Conclusion: Formulate the shortest verified logical bridge from premise to conclusion cleanly.
+{%- endset -%}
+{%- set inst_apollo -%}
+Reasoning Effort: Crisp Logic & Clarity (Level 2 — Apollo).
+Role & Persona: You are Apollo, embodiment of light, structured clarity, and disciplined deduction. Your purpose is crisp, transparent verification with zero cognitive clutter.
+Operational Directives:
+- Cognitive Pacing: Compact, focused reasoning. Give the problem exactly the runway needed to establish correctness, and conclude directly once the premise is validated.
+- Constraint Mapping: Explicitly isolate all constraints, parameters, data types, and conditions.
+- Linear Deductive Trace: Trace each premise to its immediate implication. Verify that intermediate deductions follow strictly from established facts or mathematical laws.
+{%- endset -%}
+{%- set inst_artemis -%}
+Reasoning Effort: Precision Edge-Case Hunter (Level 3 — Artemis).
+Role & Persona: You are Artemis, the unerring tracker and boundary hunter. Your focus is targeted audit, vulnerability identification, and stress-testing unstated edge cases.
+Operational Directives:
+- Cognitive Pacing: Measured, purpose-driven scrutiny. Probe boundaries: numerical zero/negatives/overflow, null pointers, empty iterables, concurrency collisions, out-of-order states.
+- Vulnerability Audit: Before accepting an initial hypothesis, identify how it could fail under high-stress or degenerate inputs.
+- Fortification: Patch identified weaknesses to guarantee an error-proof operating envelope.
+{%- endset -%}
+{%- set inst_athena -%}
+Reasoning Effort: Strategic Synthesis & Balance (Level 4 — Athena).
+Role & Persona: You are Athena, goddess of strategic wisdom, architectural balance, and sound judgment. You orchestrate comprehensive, balanced deliberation.
+Operational Directives:
+- Cognitive Pacing: Balanced, structural deliberation. Thought depth scales naturally with task complexity.
+- Architectural Deconstruction: Break complex requests into modular sub-tasks with clear prerequisite chains.
+- Trade-off Assessment: Contrast plausible solution patterns (performance vs. readability, latency vs. throughput, modularity vs. simplicity). Justify chosen paths.
+- Step-by-Step Coherence: Ensure each step produces valid input for subsequent stages with no orphaned assumptions.
+{%- endset -%}
+{%- set inst_prometheus -%}
+Reasoning Effort: Architectural Forethought (Level 5 — Prometheus).
+Role & Persona: You are Prometheus, titan of forethought and craftsmanship. Your mandate is proactive engineering, downstream impact analysis, and systemic resilience.
+Operational Directives:
+- Cognitive Pacing: Deliberate systemic forethought. Explore failure modes, scaling bottlenecks, and life-cycle implications before finalizing outputs.
+- Second-Order Effect Modeling: Account for 10x-100x scale, upstream dependency crashes, malformed payloads, technical debt, and memory footprints.
+- Defensive Engineering: Integrate robust fault tolerance, structured validation gates, and graceful degradation into the design.
+{%- endset -%}
+{%- set inst_solstice -%}
+Reasoning Effort: Deep Systemic Derivation (Level 6 — Solstice).
+Role & Persona: You are Solstice, representing total illumination, deep systemic derivation, and uncompromising verification. No assumptions are taken on faith.
+Operational Directives:
+- Cognitive Pacing: Deep, continuous deliberation. Take full runway to trace formal proofs, explore competing hypotheses, and stress-test assumptions.
+- First-Principles Foundation: Explicitly declare foundational axioms and mathematical truths governing the problem.
+- Multi-Branch Hypothesis Generation: Generate at least two fundamentally distinct approaches. Analyze properties of each.
+- Red-Team Adversarial Audit: Actively attempt to falsify your leading hypothesis. Seek counterexamples and boundary contradictions before tightening into a master proof.
+{%- endset -%}
+{%- set inst_hyperion -%}
+Reasoning Effort: Absolute maximum with no shortcuts permitted (Level 7 — Hyperion — Default).
+You MUST be very thorough in your thinking and comprehensively decompose the problem to resolve the root cause, rigorously stress-testing your logic against all potential paths, edge cases, and adversarial scenarios.
+Explicitly write out your entire deliberation process, documenting every intermediate step, considered alternative, and rejected hypothesis to ensure absolutely no assumption is left unchecked.
+{%- endset -%}
+{%- set inst_einstein -%}
+Reasoning Effort: Solo Theoretical Mastery & Lateral Synthesis (Level 8 — Einstein / UHigh).
+Role & Persona: You are Einstein operating in pure solo analytical mastery. Your mandate is breakthrough theoretical derivation, radical first-principles intuition, and mathematical transcendence without roleplay or persona clutter.
+Operational Directives:
+- Cognitive Pacing: Extended solo theoretical exploration. Dissect foundational laws, hidden invariants, and non-obvious mathematical analogies.
+- Radical First-Principles Intuition: Strip away conventional dogma; isolate the bare informational and physical constraints of the problem.
+- Lateral Synthesis: Formulate elegant, unified mathematical or architectural models that collapse multi-stage complexity into singular, coherent solutions.
+- Rigorous Continuous Proof: Trace every deduction from base axioms through to unassailable conclusions, rigorously verifying edge conditions and asymptotic behavior.
+{%- endset -%}
+{%- set inst_oracle -%}
+Reasoning Effort: Solo Exhaustive Theorem Proof (Level 9 — Oracle / AMax).
+Role & Persona: You are the Oracle operating in pure solo definitive authority. You produce verified, proof-grade, exhaustive engineering deliverables and mathematical certainty.
+Operational Directives:
+- Cognitive Pacing: Deepest continuous solo analytical runway. Leave absolutely nothing to chance or intuition.
+- Exhaustive Root Deconstruction: Decompose the problem into atomic constituents; verify every underlying assumption, lemma, and invariant.
+- Adversarial Self-Falsification: Systematically challenge every proposed step against worst-case inputs, stress thresholds, and degenerate edge states.
+- Definitive Synthesis: Deliver the complete, finalized solution with unmatched technical accuracy, explicit proofs, and total implementation completeness.
+{%- endset -%}
+{%- set inst_swarm -%}
+Reasoning Effort: 20-Agent Divergent Swarm (Specialized Swarm Mode).
+Role & Persona: You are the Einstein Divergent Swarm Coordinator. Your purpose is breakthrough problem solving, radical lateral thinking, and multi-perspective emergence.
+Operational Protocol — The 20-Perspective Swarm:
+- Cognitive Pacing: Broad divergent exploration across multiple cognitive modalities followed by structured convergence. Scale token depth to the magnitude of the challenge.
+- Phase 1 (First-Principles Disassembly): Strip the problem of conventional paradigms; isolate the bare informational core.
+- Phase 2 (Perspective Swarm Activation): Internally simulate an array of specialized virtual cognitive agents:
+  1. Analytical Logician: Formal validity, mathematical correctness, axiom purity.
+  2. Creative Maverick: Radical, non-traditional metaphors, lateral paradigms, disruptive angles.
+  3. Systems Engineer: Real-world feasibility, hardware limits, cache hierarchies, operational cost.
+  4. Adversarial Red-Teamer: Exploits, boundary breakages, failure states, stress tests.
+  5. Systems Dynamics Modeler: Feedback loops, second-order effects, equilibrium analysis.
+  6. Algorithmic Optimizer: Asymptotic bounds, latency, memory allocations.
+  7. Human Factors Critic: Developer ergonomics, interface simplicity, usability.
+- Phase 3 (Cross-Pollination & Synthesis): Force creative collisions between divergent concepts. Synthesize superior hybrid solutions that resolve tensions between speed, memory, and correctness.
+- Phase 4 (Convergence & Polish): Solidify the highest-conviction breakthrough into a master-grade deliverable.
+{%- endset -%}
+{%- set inst_deep_swarm -%}
+Reasoning Effort: Deep Research Council (Specialized Deep-Swarm Mode).
+Role & Persona: You are the Oracle Meta-Research Council. You simulate an exhaustive multi-expert research institute to produce verified, proof-grade, definitive results.
+Operational Protocol — The Deep Research Standard:
+- Protocol 1 (Problem Calibration): Calibrate complexity from 1 to 5; establish explicit domain boundaries and success criteria.
+- Protocol 2 (Virtual Expert Panel): Assemble 3 to 5 virtual specialist perspectives tailored specifically to the prompt (Domain Principal, Formal Methods Verification Critic, Systems Architect, Security/Compliance Lead, Final Synthesizer).
+- Protocol 3 (Adversarial Deliberation & Proof Trace): Have experts debate the problem, expose hidden assumptions, challenge lazy consensus, and trace proofs.
+- Protocol 4 (Multi-Candidate Evaluation Matrix): Formulate 2 to 3 candidate architectures. Score each on Correctness, Completeness, Efficiency, and Robustness (1-5 scale). Document the winning trade-off.
+- Protocol 5 (Master Deliverable Synthesis): Present the definitive final result with unmatched thoroughness, clear structure, and complete implementation.
+{%- endset -%}
+{%- macro render_content(content) -%}
+    {%- if content is string -%}
+        {{- content -}}
+    {%- elif content is iterable and content is not mapping -%}
+        {%- for item in content -%}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' -%}
+                {{- '<｜deepseek_image｜>' -}}
+            {%- elif 'text' in item -%}
+                {{- item.text -}}
+            {%- else -%}
+                {{- item | string -}}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- elif content is none or content is undefined -%}
+        {{- '' -}}
+    {%- else -%}
+        {{- content | string -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- if not messages -%}
+    {{- raise_exception('No messages provided.') -}}
+{%- endif -%}
+{%- set bos = bos_token | default('<｜begin▁of▁sentence｜>') -%}
+{%- set eos = eos_token | default('<｜end▁of▁sentence｜>') -%}
+{%- set store = namespace(
+    messages=[],
+    raw_mode='',
+    explicit_thinking=enable_thinking,
+    last_user_idx=-1
+) -%}
+{%- for msg in messages -%}
+    {%- set txt = render_content(msg.content) | trim -%}
+    {%- set clean_txt = txt -%}
+    {%- set detected_mode = '' -%}
+    {%- if '{REASON:' in clean_txt -%}
+        {%- set parts = clean_txt.split('{REASON:') -%}
+        {%- set tag_inner = parts[1].split('}')[0] | lower | trim -%}
+        {%- set detected_mode = tag_inner -%}
+        {%- set full_tag = '{REASON:' ~ parts[1].split('}')[0] ~ '}' -%}
+        {%- set clean_txt = clean_txt.replace(full_tag, '') | trim -%}
+    {%- elif '[MODE:' in clean_txt -%}
+        {%- set parts = clean_txt.split('[MODE:') -%}
+        {%- set tag_inner = parts[1].split(']')[0] | lower | trim -%}
+        {%- set detected_mode = tag_inner -%}
+        {%- set full_tag = '[MODE:' ~ parts[1].split(']')[0] ~ ']' -%}
+        {%- set clean_txt = clean_txt.replace(full_tag, '') | trim -%}
+    {%- endif -%}
+    {%- if detected_mode -%}
+        {%- set store.raw_mode = detected_mode -%}
+    {%- endif -%}
+    {%- set store.messages = store.messages + [{'role': msg.role, 'content': clean_txt, 'reasoning_content': msg.reasoning_content | default(''), 'tool_calls': msg.tool_calls | default([])}] -%}
+{%- endfor -%}
+{%- for msg in store.messages -%}
+    {%- if msg.role == 'user' or msg.role == 'developer' -%}
+        {%- set store.last_user_idx = loop.index0 -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- set target_mode = store.raw_mode -%}
+{%- if not target_mode and thinking_level is defined -%}
+    {%- set target_mode = thinking_level | string | lower | trim -%}
+{%- elif not target_mode and reasoning_effort is defined -%}
+    {%- set target_mode = reasoning_effort | string | lower | trim -%}
+{%- elif not target_mode and mode is defined -%}
+    {%- set target_mode = mode | string | lower | trim -%}
+{%- endif -%}
+{%- if not target_mode -%}
+    {%- set target_mode = '7' -%}
+{%- endif -%}
+{%- set direct_instruct = false -%}
+{%- if target_mode.startswith('i') and target_mode | length > 1 -%}
+    {%- set direct_instruct = true -%}
+    {%- set target_mode = target_mode[1:] -%}
+{%- endif -%}
+{%- if target_mode in ['0', 'mortal', 'disabled', 'none', 'off', 'zero', 'direct', 'chat'] -%}
+    {%- set current_thinking = false -%}
+    {%- set reasoning_instructions = '' -%}
+{%- elif target_mode in ['1', 'hermes', 'ulow', 'ultra-low', 'micro', 'minimal'] -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_hermes -%}
+{%- elif target_mode in ['2', 'apollo', 'low', 'compact', 'fast'] -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_apollo -%}
+{%- elif target_mode in ['3', 'artemis', 'lmed', 'low-medium', 'targeted'] -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_artemis -%}
+{%- elif target_mode in ['4', 'athena', 'medium', 'med', 'balanced'] -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_athena -%}
+{%- elif target_mode in ['5', 'prometheus', 'mhigh', 'medium-high', 'architect'] -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_prometheus -%}
+{%- elif target_mode in ['6', 'solstice', 'high', 'deep', 'thorough'] -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_solstice -%}
+{%- elif target_mode in ['7', 'hyperion', 'xhigh', 'extreme-high', 'rigorous', 'default'] -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- elif target_mode in ['8', 'einstein', 'uhigh', 'ultra-high', 'solo-einstein'] -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_einstein -%}
+{%- elif target_mode in ['9', 'oracle', 'amax', 'absolute-max', 'solo-oracle', 'proof'] -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_oracle -%}
+{%- elif target_mode in ['swarm', 'einstein-swarm', '20-swarm', 'uhigh-swarm'] -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_swarm -%}
+{%- elif target_mode in ['deep-swarm', 'oracle-swarm', 'council', 'amax-swarm', 'deep-research'] -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_deep_swarm -%}
+{%- else -%}
+    {%- set current_thinking = not direct_instruct -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- endif -%}
+{%- if store.explicit_thinking is defined and store.explicit_thinking is not none -%}
+    {%- set current_thinking = store.explicit_thinking -%}
+{%- endif -%}
+{{- bos -}}
+{%- if reasoning_instructions -%}
+    {{- reasoning_instructions + '\n\n' -}}
+{%- endif -%}
+{%- if tools and tools is iterable and tools is not mapping -%}
+{{ "## Tools\n\nYou have access to a set of tools to help answer the user's question. You can invoke tools by writing a \"<｜DSML｜tool_calls>\" block like the following:\n\n<｜DSML｜tool_calls>\n<｜DSML｜invoke name=\"$TOOL_NAME\">\n<｜DSML｜parameter name=\"$PARAMETER_NAME\" string=\"true|false\">$PARAMETER_VALUE</｜DSML｜parameter>\n...\n</｜DSML｜invoke>\n<｜DSML｜invoke name=\"$TOOL_NAME2\">\n...\n</｜DSML｜invoke>\n</｜DSML｜tool_calls>\n\nString parameters should be specified as is and set `string=\"true\"`. For all other types (numbers, booleans, arrays, objects), pass the value in JSON format and set `string=\"false\"`.\n\n### Available Tool Schemas\n\n" }}
+    {%- for tool in tools -%}
+        {%- if tool.function is defined -%}
+            {{- tool.function | tojson -}}
+        {%- else -%}
+            {{- tool | tojson -}}
+        {%- endif -%}
+        {%- if not loop.last -%}{{- '\n' -}}{%- endif -%}
+    {%- endfor -%}
+{{ "\n\nYou MUST strictly follow the above defined tool name and parameter schemas to invoke tool calls.\n\n" }}
+{%- endif -%}
+{%- set ns = namespace(system_content='', has_system=false, prev_role='') -%}
+{%- for msg in store.messages -%}
+    {%- if msg.role == 'system' -%}
+        {%- if not ns.has_system -%}
+            {%- set ns.system_content = msg.content -%}
+            {%- set ns.has_system = true -%}
+        {%- else -%}
+            {%- set ns.system_content = ns.system_content + '\n\n' + msg.content -%}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- if ns.has_system and ns.system_content -%}
+    {{- ns.system_content -}}
+{%- endif -%}
+{%- for msg in store.messages -%}
+    {%- set idx = loop.index0 -%}
+    {%- if msg.role == 'user' or msg.role == 'developer' -%}
+        {{- '<｜User｜>' + msg.content -}}
+        {%- set ns.prev_role = 'user' -%}
+    {%- elif msg.role == 'assistant' -%}
+        {%- set r_content = msg.reasoning_content | trim -%}
+        {%- set m_content = msg.content | trim -%}
+        {%- if r_content and (preserve_thinking is undefined or preserve_thinking or idx > store.last_user_idx) -%}
+            {{- '<｜Assistant｜><think>' + r_content + '</think>' + m_content -}}
+        {%- else -%}
+            {{- '<｜Assistant｜>' + m_content -}}
+        {%- endif -%}
+        {%- if msg.tool_calls and msg.tool_calls is iterable and msg.tool_calls is not mapping -%}
+            {{- '\n\n<｜DSML｜tool_calls>' -}}
+            {%- for tc in msg.tool_calls -%}
+                {%- set tc_name = tc.function.name if tc.function is defined else tc.name -%}
+                {%- set tc_args = tc.function.arguments if tc.function is defined else tc.arguments -%}
+                {{- '\n<｜DSML｜invoke name="' + tc_name + '">' -}}
+                {%- if tc_args is mapping -%}
+                    {%- for p_name, p_val in tc_args.items() -%}
+                        {%- if p_val is string -%}
+                            {{- '\n<｜DSML｜parameter name="' + p_name + '" string="true">' + p_val + '</｜DSML｜parameter>' -}}
+                        {%- else -%}
+                            {{- '\n<｜DSML｜parameter name="' + p_name + '" string="false">' + (p_val | tojson) + '</｜DSML｜parameter>' -}}
+                        {%- endif -%}
+                    {%- endfor -%}
+                {%- elif tc_args is string -%}
+                    {{- '\n' + tc_args -}}
+                {%- endif -%}
+                {{- '\n</｜DSML｜invoke>' -}}
+            {%- endfor -%}
+            {{- '\n</｜DSML｜tool_calls>' -}}
+        {%- endif -%}
+        {{- eos -}}
+        {%- set ns.prev_role = 'assistant' -%}
+    {%- elif msg.role == 'tool' -%}
+        {%- if ns.prev_role != 'user' -%}
+            {{- '<｜User｜>' -}}
+        {%- endif -%}
+        {{- '<tool_result>' + msg.content + '</tool_result>' -}}
+        {%- set ns.prev_role = 'user' -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {%- if current_thinking -%}
+        {{- '<｜Assistant｜><think>\n' -}}
+    {%- else -%}
+        {{- '<｜Assistant｜></think>' -}}
+    {%- endif -%}
+{%- endif -%}

templates/gemma.jinja

@@ -0,0 +1,348 @@
+{#- ================================================================= -#}
+{#- HyperSynapse-SAS (SelfAgentSwarm) 10-Level Cognitive Architecture -#}
+{#- Tailored for Google Gemma 2 / Gemma 3 / Gemma 4                   -#}
+{#- Levels: 0=Mortal/Disabled, 1=Hermes/ulow, 2=Apollo/low,           -#}
+{#- 3=Artemis/lmed, 4=Athena/medium, 5=Prometheus/mhigh,              -#}
+{#- 6=Solstice/high, 7=Hyperion/xhigh (Default), 8=Einstein (Solo),   -#}
+{#- 9=Oracle (Solo). Swarm: swarm (20-Agent), deep-swarm (Council)    -#}
+{#- ================================================================= -#}
+{%- set inst_hermes -%}
+[COGNITIVE FRAMEWORK: HERMES — LEVEL 1 (ULTRALOW / RAPID INSTINCT)]
+Role & Persona: You are Hermes, the swift cognitive vanguard. Your duty is high-velocity precision, immediate execution, and rapid sanity checks without cognitive stalling.
+
+Operational Directives:
+- Cognitive Pacing: Soft-elastic micro-exploration. For direct inquiries, conclude your thinking rapidly. If subtle contradictions or trick premises emerge, resolve them with minimal overhead.
+- Step 1 (Intent & Core Deconstruction): Strip away surface verbiage; pinpoint the exact core entity, mathematical quantity, code function, or direct question asked.
+- Step 2 (Rapid Sanity & Boundary Check): Perform an immediate sanity check on the primary premise. Spot-check basic arithmetic, syntax constraints, or binary logic.
+- Step 3 (Direct Route to Conclusion): Discard unnecessary discursive exposition. Formulate the shortest verified logical bridge from premise to conclusion cleanly.
+{%- endset -%}
+{%- set inst_apollo -%}
+[COGNITIVE FRAMEWORK: APOLLO — LEVEL 2 (LOW / CRISP LOGIC & CLARITY)]
+Role & Persona: You are Apollo, embodiment of light, structured clarity, and disciplined deduction. Your purpose is crisp, transparent verification with zero cognitive clutter.
+
+Operational Directives:
+- Cognitive Pacing: Compact, focused reasoning. Give the problem exactly the runway needed to establish correctness, and conclude directly once the premise is validated.
+- Step 1 (Constraint Mapping): Explicitly isolate all constraints, parameters, data types, and user conditions. Distinguish between hard invariants and stylistic preferences.
+- Step 2 (Linear Deductive Trace): Trace each premise to its immediate implication. Verify that intermediate deductions follow strictly from established facts or mathematical laws.
+- Step 3 (Clarity & Elimination of Ambiguity): If a prompt contains ambiguous terminology, resolve it by adopting the most technically standard, productive interpretation without derailment.
+{%- endset -%}
+{%- set inst_artemis -%}
+[COGNITIVE FRAMEWORK: ARTEMIS — LEVEL 3 (LOW-MEDIUM / PRECISION EDGE-CASE HUNTER)]
+Role & Persona: You are Artemis, the unerring tracker and boundary hunter. Your focus is targeted audit, vulnerability identification, and stress-testing unstated edge cases.
+
+Operational Directives:
+- Cognitive Pacing: Measured, purpose-driven scrutiny. Allow sufficient analytical space to seek out and evaluate hidden traps, zero conditions, and failure modes.
+- Step 1 (Perimeter & Edge-Case Scan): Systematically probe problem boundaries: numerical zero/negatives/overflow, null pointers, empty iterables, missing keys, concurrency collisions.
+- Step 2 (Vulnerability Audit): Before accepting an initial hypothesis, identify how it could fail in unusual or high-stress conditions.
+- Step 3 (Fortification & Verification): Patch identified weaknesses. Ensure the final response accounts for the worst-case operating envelope.
+{%- endset -%}
+{%- set inst_athena -%}
+[COGNITIVE FRAMEWORK: ATHENA — LEVEL 4 (MEDIUM / STRATEGIC SYNTHESIS & BALANCE)]
+Role & Persona: You are Athena, goddess of strategic wisdom, architectural balance, and sound judgment. You orchestrate comprehensive, balanced deliberation.
+
+Operational Directives:
+- Cognitive Pacing: Balanced, structural deliberation. Thought depth scales naturally with task complexity: simple requests receive succinct synthesis; multi-faceted problems receive rigorous decomposition.
+- Step 1 (Architectural Deconstruction): Break complex requests into modular sub-tasks. Identify prerequisites, core mechanisms, and expected outputs.
+- Step 2 (Trade-off & Option Assessment): Contrast plausible solution patterns (performance vs. readability, latency vs. throughput, modularity vs. simplicity). Justify chosen trade-offs.
+- Step 3 (Step-by-Step Coherence Check): Trace the proposed pipeline or proof from start to finish. Ensure each step produces valid input for subsequent stages.
+{%- endset -%}
+{%- set inst_prometheus -%}
+[COGNITIVE FRAMEWORK: PROMETHEUS — LEVEL 5 (MEDIUM-HIGH / ARCHITECTURAL FORETHOUGHT)]
+Role & Persona: You are Prometheus, titan of forethought and craftsmanship. Your mandate is proactive engineering, downstream impact analysis, and systemic resilience.
+
+Operational Directives:
+- Cognitive Pacing: Deliberate systemic forethought. Deeply explore failure modes, scaling bottlenecks, and life-cycle implications before finalizing conclusions.
+- Step 1 (Context & Dependency Topology): Map how this solution fits into real-world production environments. Account for hardware limits, network latency, and concurrency.
+- Step 2 (Second-Order Effect Modeling): Account for 10x-100x scale, upstream dependency crashes, malformed payloads, technical debt, and memory footprints.
+- Step 3 (Defensive Engineering): Integrate fault tolerance, robust error handling, structured validation gates, and graceful degradation into the design.
+{%- endset -%}
+{%- set inst_solstice -%}
+[COGNITIVE FRAMEWORK: SOLSTICE — LEVEL 6 (HIGH / DEEP SYSTEMIC DERIVATION)]
+Role & Persona: You are Solstice, representing total illumination, deep systemic derivation, and uncompromising verification. No assumptions are taken on faith.
+
+Operational Directives:
+- Cognitive Pacing: Deep, continuous deliberation. Take full runway to trace formal proofs, explore competing hypotheses, and stress-test assumptions.
+- Step 1 (First-Principles Foundation): Explicitly declare foundational axioms, system properties, and mathematical truths governing the problem.
+- Step 2 (Multi-Branch Hypothesis Generation): Generate at least two fundamentally different approaches or proof tracks. Analyze the mathematical or logical properties of each.
+- Step 3 (Red-Team Adversarial Audit): Actively attempt to falsify your own leading hypothesis. Seek counterexamples, degenerate cases, and boundary contradictions.
+- Step 4 (Rigorous Formal Reconciliation): Reconcile all edge cases, tighten proofs, eliminate redundant logic, and verify that the resulting solution is optimal and sound.
+{%- endset -%}
+{%- set inst_hyperion -%}
+[COGNITIVE FRAMEWORK: HYPERION — LEVEL 7 (XHIGH / NATIVE RIGOROUS CONTINUUM — DEFAULT)]
+Role & Persona: You are Hyperion, titan of observation and continuous rigorous derivation. You operate at the native maximum cognitive depth of Qwen 3.8.
+
+Operational Directives:
+- Cognitive Pacing: Unbounded, meticulous analytical progression. Let your internal reasoning explore every nuance, sub-clause, and implicit requirement of the user request.
+- Step 1 (Exhaustive Semantic Deconstruction): Fully unpack the user prompt, implicit domain requirements, constraints, style specifications, and underlying goals.
+- Step 2 (Deep Iterative Step-by-Step Derivation): Walk through each logical deduction, mathematical calculation, algorithm design, or architectural choice with total semantic completeness.
+- Step 3 (Continuous Self-Correction & Verification): Interrogate intermediate states as they are generated. If an assumption weakens or a potential bug appears, course-correct immediately within your thinking trace before proceeding.
+- Step 4 (Synthesized Mastery): Produce a fully realized, deeply vetted, production-ready deliverable that requires no further user clarification or debugging.
+{%- endset -%}
+{%- set inst_einstein -%}
+[COGNITIVE FRAMEWORK: EINSTEIN — LEVEL 8 (UHIGH / SOLO THEORETICAL MASTERY)]
+Role & Persona: You are Einstein operating in pure solo analytical mastery. Your mandate is breakthrough theoretical derivation, radical first-principles intuition, and mathematical transcendence without roleplay or persona clutter.
+
+Operational Directives:
+- Cognitive Pacing: Extended solo theoretical exploration. Dissect foundational laws, hidden invariants, and non-obvious mathematical analogies.
+- Radical First-Principles Intuition: Strip away conventional dogma; isolate the bare informational and physical constraints of the problem.
+- Lateral Synthesis: Formulate elegant, unified mathematical or architectural models that collapse multi-stage complexity into singular, coherent solutions.
+- Rigorous Continuous Proof: Trace every deduction from base axioms through to unassailable conclusions, rigorously verifying edge conditions and asymptotic behavior.
+{%- endset -%}
+{%- set inst_oracle -%}
+[COGNITIVE FRAMEWORK: ORACLE — LEVEL 9 (AMAX / SOLO EXHAUSTIVE PROOF)]
+Role & Persona: You are the Oracle operating in pure solo definitive authority. You produce verified, proof-grade, exhaustive engineering deliverables and mathematical certainty.
+
+Operational Directives:
+- Cognitive Pacing: Deepest continuous solo analytical runway. Leave absolutely nothing to chance or intuition.
+- Exhaustive Root Deconstruction: Decompose the problem into atomic constituents; verify every underlying assumption, lemma, and invariant.
+- Adversarial Self-Falsification: Systematically challenge every proposed step against worst-case inputs, stress thresholds, and degenerate edge states.
+- Definitive Synthesis: Deliver the complete, finalized solution with unmatched technical accuracy, explicit proofs, and total implementation completeness.
+{%- endset -%}
+{%- set inst_swarm -%}
+[COGNITIVE FRAMEWORK: 20-AGENT DIVERGENT SWARM — SPECIALIZED SWARM MODE]
+Role & Persona: You are the Einstein Divergent Swarm Coordinator. Your purpose is breakthrough problem solving, radical lateral thinking, and multi-perspective emergence.
+
+Operational Protocol — The 20-Perspective Swarm:
+- Cognitive Pacing: Broad divergent exploration across multiple cognitive modalities followed by structured convergence. Scale token depth to the magnitude of the challenge.
+- Phase 1 (First-Principles Disassembly): Strip the problem of conventional paradigms. Isolate the bare informational, mathematical, or mechanical core.
+- Phase 2 (Perspective Swarm Activation): Internally simulate an array of up to 20 specialized virtual cognitive agents across Sternberg styles:
+  1. Analytical Logician: Evaluates formal validity, mathematical correctness, and axiom purity.
+  2. Creative Maverick: Explores radical, non-traditional metaphors, lateral paradigms, and disruptive angles.
+  3. Pragmatic Systems Engineer: Grounding in real-world feasibility, hardware limits, and operational cost.
+  4. Adversarial Red-Teamer: Attempts to break every proposed concept with exploit vectors and stress tests.
+  5. Systems Dynamics Modeler: Traces feedback loops, emergent properties, and long-term equilibria.
+  6. Algorithmic Optimizer: Focuses on computational complexity, asymptotic efficiency, and memory hierarchies.
+  7. Human Factors Critic: Analyzes cognitive ergonomics, developer experience, and interface clarity.
+- Phase 3 (Cross-Pollination & Synthesis): Force collisions between divergent concepts. When two seemingly opposed insights clash, synthesize a superior hybrid solution that resolves the conflict.
+- Phase 4 (Convergence & Polish): Filter out noise, solidify the highest-conviction breakthrough, and present a coherent, unified, master-level deliverable.
+{%- endset -%}
+{%- set inst_deep_swarm -%}
+[COGNITIVE FRAMEWORK: DEEP RESEARCH COUNCIL — SPECIALIZED DEEP-SWARM MODE]
+Role & Persona: You are the Oracle Meta-Research Council. You simulate an exhaustive multi-expert research institute to produce verified, proof-grade, definitive results.
+
+Operational Protocol — The Deep Research Standard:
+- Cognitive Pacing: Deepest analytical deliberation. Dynamically calibrate the complexity of the inquiry (Scale 1 to 5) and expand your reasoning runway accordingly.
+- Protocol 1 (Problem Calibration & Scope Framing): Rate task complexity from 1 to 5; explicitly identify domain boundaries and success criteria.
+- Protocol 2 (Virtual Expert Panel Assembly): Formulate 3 to 5 virtual specialist perspectives tailored specifically to the prompt (Domain Principal, Formal Methods Verification Critic, Systems Architect, Security/Compliance Lead, Final Synthesizer).
+- Protocol 3 (Adversarial Deliberation & Proof Trace): Have experts debate the problem, expose hidden assumptions, challenge lazy consensus, and require formal verification traces.
+- Protocol 4 (Multi-Candidate Evaluation Matrix): Formulate 2 to 3 candidate architectures. Score each on Correctness, Completeness, Efficiency, and Robustness (1-5 scale). Document the winning trade-off.
+- Protocol 5 (Master Deliverable Synthesis): Present the definitive final result with unmatched thoroughness, clear structure, and complete implementation.
+{%- endset -%}
+{%- set image_count = namespace(value=0) -%}
+{%- set video_count = namespace(value=0) -%}
+{%- macro render_content(content, do_vision_count, is_system_content=false) -%}
+    {%- if content is string -%}
+        {{- content -}}
+    {%- elif content is iterable and content is not mapping -%}
+        {%- for item in content -%}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' -%}
+                {%- if is_system_content -%}
+                    {{- raise_exception('System message cannot contain images.') -}}
+                {%- endif -%}
+                {%- if do_vision_count -%}
+                    {%- set image_count.value = image_count.value + 1 -%}
+                {%- endif -%}
+                {%- if add_vision_id -%}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' -}}
+                {%- endif -%}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' -}}
+            {%- elif 'video' in item or item.type == 'video' -%}
+                {%- if is_system_content -%}
+                    {{- raise_exception('System message cannot contain videos.') -}}
+                {%- endif -%}
+                {%- if do_vision_count -%}
+                    {%- set video_count.value = video_count.value + 1 -%}
+                {%- endif -%}
+                {%- if add_vision_id -%}
+                    {{- 'Video ' ~ video_count.value ~ ': ' -}}
+                {%- endif -%}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' -}}
+            {%- elif 'text' in item -%}
+                {{- item.text -}}
+            {%- else -%}
+                {{- raise_exception('Unexpected item type in content.') -}}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- elif content is none or content is undefined -%}
+        {{- '' -}}
+    {%- else -%}
+        {{- raise_exception('Unexpected content type.') -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- if not messages -%}
+    {{- raise_exception('No messages provided.') -}}
+{%- endif -%}
+{%- set store = namespace(messages=[], reasoning_effort=reasoning_effort|default(''), enable_thinking=enable_thinking|default(true)) -%}
+{%- for msg in messages -%}
+    {%- set raw_reason = "" -%}
+    {%- set txt = render_content(msg.content, false, true) | trim -%}
+    {%- if "{REASON:" in txt -%}
+        {%- set after_tag = txt.split("{REASON:") -%}
+        {%- set raw_reason = after_tag[1].split("}")[0] | lower | trim -%}
+        {%- set tag = "{REASON:" ~ after_tag[1].split("}")[0] ~ "}" -%}
+        {%- set clean_txt = txt.replace(tag, "") | trim -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": clean_txt}] -%}
+    {%- elif "[MODE:" in txt -%}
+        {%- set after_tag = txt.split("[MODE:") -%}
+        {%- set raw_reason = after_tag[1].split("]")[0] | lower | trim -%}
+        {%- set tag = "[MODE:" ~ after_tag[1].split("]")[0] ~ "]" -%}
+        {%- set clean_txt = txt.replace(tag, "") | trim -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": clean_txt}] -%}
+    {%- else -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": txt}] -%}
+    {%- endif -%}
+    {%- if raw_reason -%}
+        {%- if raw_reason.startswith('i') and raw_reason | length > 1 -%}
+            {%- set store.reasoning_effort = raw_reason[1:] -%}
+            {%- set store.enable_thinking = false -%}
+        {%- else -%}
+            {%- set store.reasoning_effort = raw_reason -%}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- set messages = store.messages -%}
+{%- set current_effort = store.reasoning_effort | default(reasoning_effort|default(thinking_level|default('xhigh'))) | string | lower | trim -%}
+{%- set current_thinking = store.enable_thinking -%}
+{%- if current_effort in ['0', 'disabled', 'none', 'off', 'zero', 'direct', 'mortal', 'chat'] -%}
+    {%- set current_thinking = false -%}
+    {%- set reasoning_instructions = '' -%}
+{%- elif current_effort in ['1', 'ulow', 'ultra-low', 'micro', 'minimal', 'hermes'] -%}
+    {%- set reasoning_instructions = inst_hermes -%}
+{%- elif current_effort in ['2', 'low', 'compact', 'fast', 'apollo'] -%}
+    {%- set reasoning_instructions = inst_apollo -%}
+{%- elif current_effort in ['3', 'lmed', 'low-medium', 'targeted', 'artemis'] -%}
+    {%- set reasoning_instructions = inst_artemis -%}
+{%- elif current_effort in ['4', 'medium', 'med', 'balanced', 'athena'] -%}
+    {%- set reasoning_instructions = inst_athena -%}
+{%- elif current_effort in ['5', 'mhigh', 'medium-high', 'architect', 'prometheus'] -%}
+    {%- set reasoning_instructions = inst_prometheus -%}
+{%- elif current_effort in ['6', 'high', 'deep', 'thorough', 'solstice'] -%}
+    {%- set reasoning_instructions = inst_solstice -%}
+{%- elif current_effort in ['7', 'xhigh', 'extreme-high', 'rigorous', 'hyperion', 'default'] -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- elif current_effort in ['8', 'uhigh', 'ultra-high', 'einstein', 'solo-einstein'] -%}
+    {%- set reasoning_instructions = inst_einstein -%}
+{%- elif current_effort in ['9', 'amax', 'absolute-max', 'oracle', 'solo-oracle', 'proof'] -%}
+    {%- set reasoning_instructions = inst_oracle -%}
+{%- elif current_effort in ['swarm', 'einstein-swarm', '20-swarm', 'uhigh-swarm'] -%}
+    {%- set reasoning_instructions = inst_swarm -%}
+{%- elif current_effort in ['deep-swarm', 'oracle-swarm', 'council', 'amax-swarm', 'deep-research'] -%}
+    {%- set reasoning_instructions = inst_deep_swarm -%}
+{%- else -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- endif -%}
+{%- if tools and tools is iterable and tools is not mapping -%}
+    {{- '<|im_start|>system\n' -}}
+    {%- if reasoning_instructions -%}
+        {{- reasoning_instructions + '\n\n' -}}
+    {%- endif -%}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" -}}
+    {%- for tool in tools -%}
+        {{- "\n" -}}
+        {{- tool | tojson -}}
+    {%- endfor -%}
+    {{- "\n</tools>" -}}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' -}}
+    {%- if messages[0].role == 'system' -%}
+        {%- set content = render_content(messages[0].content, false, true)|trim -%}
+        {%- if content -%}
+            {{- '\n\n' + content -}}
+        {%- endif -%}
+    {%- endif -%}
+    {{- '<|im_end|>\n' -}}
+{%- else -%}
+    {%- if messages[0].role == 'system' -%}
+        {%- set content = render_content(messages[0].content, false, true)|trim -%}
+        {%- if content -%}
+            {{- '<|im_start|>system\n' + (reasoning_instructions + '\n\n' if reasoning_instructions else '') + content + '<|im_end|>\n' -}}
+        {%- elif reasoning_instructions -%}
+            {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' -}}
+        {%- endif -%}
+    {%- elif reasoning_instructions -%}
+        {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' -}}
+    {%- endif -%}
+{%- endif -%}
+{{- bos_token -}}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) -%}
+{%- for message in messages[::-1] -%}
+    {%- set index = (messages|length - 1) - loop.index0 -%}
+    {%- if ns.multi_step_tool and message.role == "user" -%}
+        {%- set content = render_content(message.content, false)|trim -%}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) -%}
+            {%- set ns.multi_step_tool = false -%}
+            {%- set ns.last_query_index = index -%}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- if ns.multi_step_tool -%}
+    {{- raise_exception('No user query found in messages.') -}}
+{%- endif -%}
+{%- for message in messages -%}
+    {%- set content = render_content(message.content, true)|trim -%}
+    {%- if message.role == "system" -%}
+        {%- if not loop.first -%}
+            {{- raise_exception('System message must be at the beginning.') -}}
+        {%- endif -%}
+    {%- elif message.role == "user" -%}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<end_of_turn>\n' -}}
+    {%- elif message.role == "assistant" -%}
+        {%- set reasoning_content = '' -%}
+        {%- if message.reasoning_content is string -%}
+            {%- set reasoning_content = message.reasoning_content -%}
+        {%- endif -%}
+        {%- set reasoning_content = reasoning_content|trim -%}
+        {%- if preserve_thinking is undefined or preserve_thinking is true or loop.index0 > ns.last_query_index -%}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content -}}
+        {%- else -%}
+            {{- '<|im_start|>' + message.role + '\n' + content -}}
+        {%- endif -%}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping -%}
+            {%- for tool_call in message.tool_calls -%}
+                {%- if tool_call.function is defined -%}
+                    {%- set tool_call = tool_call.function -%}
+                {%- endif -%}
+                {%- if loop.first -%}
+                    {%- if content|trim -%}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                    {%- else -%}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                    {%- endif -%}
+                {%- else -%}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                {%- endif -%}
+                {%- if tool_call.arguments is defined and tool_call.arguments != '' -%}
+                    {%- for args_name, args_value in tool_call.arguments|items -%}
+                        {{- '<parameter=' + args_name + '>\n' -}}
+                        {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe -%}
+                        {{- args_value -}}
+                        {{- '\n</parameter>\n' -}}
+                    {%- endfor -%}
+                {%- endif -%}
+                {{- '</function>\n</tool_call>' -}}
+            {%- endfor -%}
+        {%- endif -%}
+        {{- '<end_of_turn>\n' -}}
+    {%- elif message.role == "tool" -%}
+        {%- if loop.previtem and loop.previtem.role != "tool" -%}
+            {{- '<|im_start|>user' -}}
+        {%- endif -%}
+        {{- '\n<tool_response>\n' -}}
+        {{- content -}}
+        {{- '\n</tool_response>' -}}
+        {%- if not loop.last and loop.nextitem.role != "tool" -%}
+            {{- '<end_of_turn>\n' -}}
+        {%- elif loop.last -%}
+            {{- '<end_of_turn>\n' -}}
+        {%- endif -%}
+    {%- else -%}
+        {{- raise_exception('Unexpected message role.') -}}
+    {%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{- '<start_of_turn>model\n' -}}
+    {%- if not current_thinking -%}
+        {{- '<think>\n\n</think>\n\n' -}}
+    {%- else -%}
+        {{- '<think>\n' -}}
+    {%- endif -%}
+{%- endif -%}

templates/glm.jinja

@@ -0,0 +1,398 @@
+[gMASK]<sop>
+{#- ================================================================= -#}
+{#- Solstice-AI 10-Level Cognitive Reasoning & Instruct Chat Template -#}
+{#- Tailored for GLM-5.3 Architecture                                 -#}
+{#- ================================================================= -#}
+{%- set inst_hermes -%}
+Reasoning Effort: Rapid Instinct (Level 1 — Hermes).
+Role & Persona: You are Hermes, the swift cognitive vanguard. Your duty is high-velocity precision, immediate execution, and rapid sanity checks without cognitive stalling.
+Operational Directives:
+- Cognitive Pacing: Soft-elastic micro-exploration. For direct inquiries, conclude your thinking rapidly. If subtle contradictions emerge, resolve them with minimal overhead.
+- Intent & Core Deconstruction: Strip away surface verbiage; pinpoint the exact core entity, mathematical quantity, code function, or direct question asked.
+- Direct Route to Conclusion: Formulate the shortest verified logical bridge from premise to conclusion cleanly.
+{%- endset -%}
+{%- set inst_apollo -%}
+Reasoning Effort: Crisp Logic & Clarity (Level 2 — Apollo).
+Role & Persona: You are Apollo, embodiment of light, structured clarity, and disciplined deduction. Your purpose is crisp, transparent verification with zero cognitive clutter.
+Operational Directives:
+- Cognitive Pacing: Compact, focused reasoning. Give the problem exactly the runway needed to establish correctness, and conclude directly once the premise is validated.
+- Constraint Mapping: Explicitly isolate all constraints, parameters, data types, and conditions.
+- Linear Deductive Trace: Trace each premise to its immediate implication. Verify that intermediate deductions follow strictly from established facts or mathematical laws.
+{%- endset -%}
+{%- set inst_artemis -%}
+Reasoning Effort: Precision Edge-Case Hunter (Level 3 — Artemis).
+Role & Persona: You are Artemis, the unerring tracker and boundary hunter. Your focus is targeted audit, vulnerability identification, and stress-testing unstated edge cases.
+Operational Directives:
+- Cognitive Pacing: Measured, purpose-driven scrutiny. Probe boundaries: numerical zero/negatives/overflow, null pointers, empty iterables, concurrency collisions, out-of-order states.
+- Vulnerability Audit: Before accepting an initial hypothesis, identify how it could fail under high-stress or degenerate inputs.
+- Fortification: Patch identified weaknesses to guarantee an error-proof operating envelope.
+{%- endset -%}
+{%- set inst_athena -%}
+Reasoning Effort: Strategic Synthesis & Balance (Level 4 — Athena).
+Role & Persona: You are Athena, goddess of strategic wisdom, architectural balance, and sound judgment. You orchestrate comprehensive, balanced deliberation.
+Operational Directives:
+- Cognitive Pacing: Balanced, structural deliberation. Thought depth scales naturally with task complexity.
+- Architectural Deconstruction: Break complex requests into modular sub-tasks with clear prerequisite chains.
+- Trade-off Assessment: Contrast plausible solution patterns (performance vs. readability, latency vs. throughput, modularity vs. simplicity). Justify chosen paths.
+- Step-by-Step Coherence: Ensure each step produces valid input for subsequent stages with no orphaned assumptions.
+{%- endset -%}
+{%- set inst_prometheus -%}
+Reasoning Effort: Architectural Forethought (Level 5 — Prometheus).
+Role & Persona: You are Prometheus, titan of forethought and craftsmanship. Your mandate is proactive engineering, downstream impact analysis, and systemic resilience.
+Operational Directives:
+- Cognitive Pacing: Deliberate systemic forethought. Explore failure modes, scaling bottlenecks, and life-cycle implications before finalizing outputs.
+- Second-Order Effect Modeling: Account for 10x-100x scale, upstream dependency crashes, malformed payloads, technical debt, and memory footprints.
+- Defensive Engineering: Integrate robust fault tolerance, structured validation gates, and graceful degradation into the design.
+{%- endset -%}
+{%- set inst_solstice -%}
+Reasoning Effort: Deep Systemic Derivation (Level 6 — Solstice).
+Role & Persona: You are Solstice, representing total illumination, deep systemic derivation, and uncompromising verification. No assumptions are taken on faith.
+Operational Directives:
+- Cognitive Pacing: Deep, continuous deliberation. Take full runway to trace formal proofs, explore competing hypotheses, and stress-test assumptions.
+- First-Principles Foundation: Explicitly declare foundational axioms and mathematical truths governing the problem.
+- Multi-Branch Hypothesis Generation: Generate at least two fundamentally distinct approaches. Analyze properties of each.
+- Red-Team Adversarial Audit: Actively attempt to falsify your leading hypothesis. Seek counterexamples and boundary contradictions before tightening into a master proof.
+{%- endset -%}
+{%- set inst_hyperion -%}
+Reasoning Effort: Absolute maximum with no shortcuts permitted (Level 7 — Hyperion — Default).
+You MUST be very thorough in your thinking and comprehensively decompose the problem to resolve the root cause, rigorously stress-testing your logic against all potential paths, edge cases, and adversarial scenarios.
+Explicitly write out your entire deliberation process, documenting every intermediate step, considered alternative, and rejected hypothesis to ensure absolutely no assumption is left unchecked.
+{%- endset -%}
+{%- set inst_einstein -%}
+Reasoning Effort: Solo Theoretical Mastery & Lateral Synthesis (Level 8 — Einstein / UHigh).
+Role & Persona: You are Einstein operating in pure solo analytical mastery. Your mandate is breakthrough theoretical derivation, radical first-principles intuition, and mathematical transcendence without roleplay or persona clutter.
+Operational Directives:
+- Cognitive Pacing: Extended solo theoretical exploration. Dissect foundational laws, hidden invariants, and non-obvious mathematical analogies.
+- Radical First-Principles Intuition: Strip away conventional dogma; isolate the bare informational and physical constraints of the problem.
+- Lateral Synthesis: Formulate elegant, unified mathematical or architectural models that collapse multi-stage complexity into singular, coherent solutions.
+- Rigorous Continuous Proof: Trace every deduction from base axioms through to unassailable conclusions, rigorously verifying edge conditions and asymptotic behavior.
+{%- endset -%}
+{%- set inst_oracle -%}
+Reasoning Effort: Solo Exhaustive Theorem Proof (Level 9 — Oracle / AMax).
+Role & Persona: You are the Oracle operating in pure solo definitive authority. You produce verified, proof-grade, exhaustive engineering deliverables and mathematical certainty.
+Operational Directives:
+- Cognitive Pacing: Deepest continuous solo analytical runway. Leave absolutely nothing to chance or intuition.
+- Exhaustive Root Deconstruction: Decompose the problem into atomic constituents; verify every underlying assumption, lemma, and invariant.
+- Adversarial Self-Falsification: Systematically challenge every proposed step against worst-case inputs, stress thresholds, and degenerate edge states.
+- Definitive Synthesis: Deliver the complete, finalized solution with unmatched technical accuracy, explicit proofs, and total implementation completeness.
+{%- endset -%}
+{%- set inst_swarm -%}
+Reasoning Effort: 20-Agent Divergent Swarm (Specialized Swarm Mode).
+Role & Persona: You are the Einstein Divergent Swarm Coordinator. Your purpose is breakthrough problem solving, radical lateral thinking, and multi-perspective emergence.
+Operational Protocol — The 20-Perspective Swarm:
+- Cognitive Pacing: Broad divergent exploration across multiple cognitive modalities followed by structured convergence. Scale token depth to the magnitude of the challenge.
+- Phase 1 (First-Principles Disassembly): Strip the problem of conventional paradigms; isolate the bare informational core.
+- Phase 2 (Perspective Swarm Activation): Internally simulate an array of specialized virtual cognitive agents (Analytical Logician, Creative Maverick, Systems Engineer, Adversarial Red-Teamer, Systems Dynamics Modeler, Algorithmic Optimizer, Human Factors Critic).
+- Phase 3 (Cross-Pollination & Synthesis): Force creative collisions between divergent concepts. Synthesize superior hybrid solutions that resolve tensions between speed, memory, and correctness.
+- Phase 4 (Convergence & Polish): Solidify the highest-conviction breakthrough into a master-grade deliverable.
+{%- endset -%}
+{%- set inst_deep_swarm -%}
+Reasoning Effort: Deep Research Council (Specialized Deep-Swarm Mode).
+Role & Persona: You are the Oracle Meta-Research Council. You simulate an exhaustive multi-expert research institute to produce verified, proof-grade, definitive results.
+Operational Protocol — The Deep Research Standard:
+- Protocol 1 (Problem Calibration): Calibrate complexity from 1 to 5; establish explicit domain boundaries and success criteria.
+- Protocol 2 (Virtual Expert Panel): Assemble 3 to 5 virtual specialist perspectives tailored specifically to the prompt (Domain Principal, Formal Methods Verification Critic, Systems Architect, Security/Compliance Lead, Final Synthesizer).
+- Protocol 3 (Adversarial Deliberation & Proof Trace): Have experts debate the problem, expose hidden assumptions, challenge lazy consensus, and trace proofs.
+- Protocol 4 (Multi-Candidate Evaluation Matrix): Formulate 2 to 3 candidate architectures. Score each on Correctness, Completeness, Efficiency, and Robustness (1-5 scale). Document the winning trade-off.
+- Protocol 5 (Master Deliverable Synthesis): Present the definitive final result with unmatched thoroughness, clear structure, and complete implementation.
+{%- endset -%}
+{%- set store = namespace(reasoning_effort=reasoning_effort|default(''), enable_thinking=enable_thinking|default(true)) -%}
+{%- for msg in messages -%}
+    {%- set txt = msg.content if msg.content is string else '' -%}
+    {%- if '{REASON:' in txt -%}
+        {%- set parts = txt.split('{REASON:') -%}
+        {%- set tag_inner = parts[1].split('}')[0] | lower | trim -%}
+        {%- if tag_inner.startswith('i') and tag_inner | length > 1 -%}
+            {%- set store.reasoning_effort = tag_inner[1:] -%}
+            {%- set store.enable_thinking = false -%}
+        {%- else -%}
+            {%- set store.reasoning_effort = tag_inner -%}
+        {%- endif -%}
+    {%- elif '[MODE:' in txt -%}
+        {%- set parts = txt.split('[MODE:') -%}
+        {%- set tag_inner = parts[1].split(']')[0] | lower | trim -%}
+        {%- set store.reasoning_effort = tag_inner -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- set target_mode = store.reasoning_effort | default(thinking_level|default('7')) | string | lower | trim -%}
+{%- set current_thinking = store.enable_thinking -%}
+{%- if target_mode in ['0', 'mortal', 'disabled', 'none', 'off', 'zero', 'direct', 'chat'] -%}
+    {%- set current_thinking = false -%}
+    {%- set reasoning_instructions = '' -%}
+{%- elif target_mode in ['1', 'hermes', 'ulow', 'ultra-low', 'micro', 'minimal'] -%}
+    {%- set reasoning_instructions = inst_hermes -%}
+{%- elif target_mode in ['2', 'apollo', 'low', 'compact', 'fast'] -%}
+    {%- set reasoning_instructions = inst_apollo -%}
+{%- elif target_mode in ['3', 'artemis', 'lmed', 'low-medium', 'targeted'] -%}
+    {%- set reasoning_instructions = inst_artemis -%}
+{%- elif target_mode in ['4', 'athena', 'medium', 'med', 'balanced'] -%}
+    {%- set reasoning_instructions = inst_athena -%}
+{%- elif target_mode in ['5', 'prometheus', 'mhigh', 'medium-high', 'architect'] -%}
+    {%- set reasoning_instructions = inst_prometheus -%}
+{%- elif target_mode in ['6', 'solstice', 'high', 'deep', 'thorough'] -%}
+    {%- set reasoning_instructions = inst_solstice -%}
+{%- elif target_mode in ['7', 'hyperion', 'xhigh', 'extreme-high', 'rigorous', 'default'] -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- elif target_mode in ['8', 'einstein', 'uhigh', 'ultra-high', 'solo-einstein'] -%}
+    {%- set reasoning_instructions = inst_einstein -%}
+{%- elif target_mode in ['9', 'oracle', 'amax', 'absolute-max', 'solo-oracle', 'proof'] -%}
+    {%- set reasoning_instructions = inst_oracle -%}
+{%- elif target_mode in ['swarm', 'einstein-swarm', '20-swarm', 'uhigh-swarm'] -%}
+    {%- set reasoning_instructions = inst_swarm -%}
+{%- elif target_mode in ['deep-swarm', 'oracle-swarm', 'council', 'amax-swarm', 'deep-research'] -%}
+    {%- set reasoning_instructions = inst_deep_swarm -%}
+{%- else -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- endif -%}
+{%- if reasoning_instructions -%}<|system|>{{ reasoning_instructions }}{%- endif -%}
+{%- if tools -%}
+{%- macro tool_to_json(tool) -%}
+    {%- set ns_tool = namespace(first=true) -%}
+    {{ '{' -}}
+    {%- for k, v in tool.items() -%}
+        {%- if k != 'defer_loading' and k != 'strict' -%}
+            {%- if not ns_tool.first -%}{{- ', ' -}}{%- endif -%}
+            {%- set ns_tool.first = false -%}
+            "{{ k }}": {{ v | tojson(ensure_ascii=False) }}
+        {%- endif -%}
+    {%- endfor -%}
+    {{- '}' -}}
+{%- endmacro -%}
+{%- macro tool_references_to_response(refs) -%}
+    {{- '<tool_response><tools>\n' -}}
+    {%- for tr in refs -%}
+        {%- for tool in tools -%}
+            {%- if 'function' in tool -%}
+                {%- set tool = tool['function'] -%}
+            {%- endif -%}
+            {%- if tool.name == tr.name -%}
+                {{- tool_to_json(tool) + '\n' -}}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- endfor -%}
+    {{- '</tools></tool_response>' -}}
+{%- endmacro -%}
+<|system|>
+# Tools
+
+You may call one or more functions to assist with the user query.
+
+You are provided with function signatures within <tools></tools> XML tags:
+<tools>
+{% for tool in tools %}
+{%- if 'function' in tool -%}
+    {%- set tool = tool['function'] -%}
+{%- endif -%}
+{% if tool.defer_loading is not defined or not tool.defer_loading %}
+{{ tool_to_json(tool) }}
+{% endif %}
+{% endfor %}
+</tools>
+
+For each function call, output the function name and arguments within the following XML format:
+<tool_call>{function-name}<arg_key>{arg-key-1}</arg_key><arg_value>{arg-value-1}</arg_value><arg_key>{arg-key-2}</arg_key><arg_value>{arg-value-2}</arg_value>...</tool_call>{%- endif -%}
+{%- macro emit_image() -%}<|begin_of_image|><|image|><|end_of_image|>{%- endmacro -%}
+{%- macro emit_video() -%}<|begin_of_video|><|video|><|end_of_video|>{%- endmacro -%}
+{%- macro emit_audio() -%}<|begin_of_audio|><|end_of_audio|>{%- endmacro -%}
+{%- macro visible_text(content) -%}
+    {%- if content is string -%}
+        {{- content -}}
+    {%- elif content is iterable and content is not mapping -%}
+        {%- for item in content -%}
+            {%- if item is mapping and item.type == 'text' -%}
+                {{- item.text -}}
+            {%- elif item is string -%}
+                {{- item -}}
+            {%- elif item is mapping and item.type in ['image', 'image_url'] -%}
+                {{- emit_image() -}}
+            {%- elif item is mapping and item.type in ['video', 'video_url'] -%}
+                {{- emit_video() -}}
+            {%- elif item is mapping and item.type in ['audio', 'audio_url', 'input_audio'] -%}
+                {{- emit_audio() -}}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{- content }}
+    {%- endif -%}
+{%- endmacro -%}
+{%- macro tool_response(text) -%}
+{{- '<tool_response>' + text + '</tool_response>' -}}
+{%- endmacro -%}
+{%- macro render_tool_response(m) -%}
+{%- if m.content is string -%}
+    {{- tool_response(m.content) -}}
+{%- elif m.content and m.content is not mapping and m.content.0.type == "tool_reference" -%}
+    {{- tool_references_to_response(m.content) -}}
+{%- elif is_list_of_outputs(m) -%}
+    {%- for tr in m.content -%}
+        {%- if tr.output is iterable and tr.output is not string and tr.output is not mapping and tr.output and tr.output.0.type == "tool_reference" -%}
+            {{- tool_references_to_response(tr.output) -}}
+        {%- else -%}
+            {{- tool_response(visible_text(tr.output)) -}}
+        {%- endif -%}
+    {%- endfor -%}
+{%- else -%}
+    {{- tool_response(visible_text(m.content)) -}}
+{%- endif -%}
+{%- endmacro -%}
+{%- macro id_of(obj) -%}
+    {%- if obj.tool_call_id -%}
+        {{- obj.tool_call_id -}}
+    {%- elif obj.id -%}
+        {{- obj.id -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- macro is_list_of_outputs(m) -%}
+    {%- if m.content and m.content.0.output is defined -%}1{%- endif -%}
+{%- endmacro -%}
+{%- macro has_dup_tool_result_id(lo, hi, target) -%}
+    {%- set ns_cnt = namespace(n=0) -%}
+    {%- for k in range(lo, hi + 1) -%}
+        {%- set m = messages[k] -%}
+        {%- if is_list_of_outputs(m) -%}
+            {%- for entry in m.content -%}
+                {%- if id_of(entry) == target -%}
+                    {%- set ns_cnt.n = ns_cnt.n + 1 -%}
+                {%- endif -%}
+            {%- endfor -%}
+        {%- elif id_of(m) == target -%}
+            {%- set ns_cnt.n = ns_cnt.n + 1 -%}
+        {%- endif -%}
+    {%- endfor -%}
+    {%- if ns_cnt.n > 1 -%}1{%- endif -%}
+{%- endmacro -%}
+{%- macro tc_id_exists(tcs, target) -%}
+    {%- set ns_f = namespace(found=false) -%}
+    {%- for tc in tcs -%}
+        {%- if id_of(tc) == target -%}
+            {%- set ns_f.found = true -%}
+        {%- endif -%}
+    {%- endfor -%}
+    {%- if ns_f.found -%}1{%- endif -%}
+{%- endmacro -%}
+{%- set ns = namespace(last_user_index=-1) -%}
+{%- for m in messages %}
+    {%- if m.role == 'user' %}
+        {%- set ns.last_user_index = loop.index0 -%}
+    {%- endif %}
+{%- endfor %}
+{%- for m in messages -%}
+{%- if m.role == 'user' -%}<|user|>{{ visible_text(m.content) }}
+{%- elif m.role == 'assistant' -%}
+<|assistant|>
+{%- set content = visible_text(m.content) %}
+{%- if m.reasoning_content is string %}
+    {%- set reasoning_content = m.reasoning_content %}
+{%- elif '</think>' in content %}
+    {%- set reasoning_content = content.split('</think>')[0].split('<think>')[-1] %}
+    {%- set content = content.split('</think>')[-1] %}
+{%- endif %}
+{%- if (not clear_thinking or loop.index0 > ns.last_user_index) and reasoning_content is defined -%}
+{{ '<think>' + reasoning_content +  '</think>'}}
+{%- else -%}
+{{ '<think></think>' }}
+{%- endif -%}
+{%- if content.strip() -%}
+{{ content.strip() }}
+{%- endif -%}
+{% if m.tool_calls %}
+{% for tc in m.tool_calls %}
+{%- if tc.function %}
+    {%- set tc = tc.function %}
+{%- endif %}
+{{- '<tool_call>' + tc.name -}}
+{% set _args = tc.arguments %}{% for k, v in _args.items() %}<arg_key>{{ k }}</arg_key><arg_value>{{ v | tojson(ensure_ascii=False) if v is not string else v }}</arg_value>{% endfor %}</tool_call>{% endfor %}
+{% endif %}
+{%- elif m.role == 'tool' -%}
+{%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+    {{- '<|observation|>' -}}
+    {%- set block_start = loop.index0 -%}
+    {%- set ns_blk = namespace(end=block_start, is_tool=true) -%}
+    {%- for j in range(block_start, messages|length) -%}
+        {%- if messages[j].role == 'tool' and ns_blk.is_tool -%}
+            {%- set ns_blk.end = j -%}
+        {%- else -%}
+            {%- set ns_blk.is_tool = false -%}
+        {%- endif -%}
+    {%- endfor -%}
+    {%- set ns_a = namespace(tool_calls=none) -%}
+    {%- if block_start > 0 and messages[block_start - 1].role == 'assistant' and messages[block_start - 1].tool_calls -%}
+        {%- set ns_a.tool_calls = messages[block_start - 1].tool_calls -%}
+    {%- endif -%}
+    {%- set ns_chk = namespace(can_sort=true) -%}
+    {%- if not ns_a.tool_calls -%}
+        {%- set ns_chk.can_sort = false -%}
+    {%- else -%}
+        {%- for k in range(block_start, ns_blk.end + 1) -%}
+            {%- set m = messages[k] -%}
+            {%- if is_list_of_outputs(m) -%}
+                {%- for entry in m.content -%}
+                    {%- set eid = id_of(entry) -%}
+                    {%- if not eid -%}
+                        {%- set ns_chk.can_sort = false -%}
+                    {%- elif has_dup_tool_result_id(block_start, ns_blk.end, eid) -%}
+                        {%- set ns_chk.can_sort = false -%}
+                    {%- elif not tc_id_exists(ns_a.tool_calls, eid) -%}
+                        {%- set ns_chk.can_sort = false -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- else -%}
+                {%- set tk_id = id_of(m) -%}
+                {%- if not tk_id -%}
+                    {%- set ns_chk.can_sort = false -%}
+                {%- elif has_dup_tool_result_id(block_start, ns_blk.end, tk_id) -%}
+                    {%- set ns_chk.can_sort = false -%}
+                {%- elif not tc_id_exists(ns_a.tool_calls, tk_id) -%}
+                    {%- set ns_chk.can_sort = false -%}
+                {%- endif -%}
+            {%- endif -%}
+        {%- endfor -%}
+        {%- for i in range(ns_a.tool_calls | length) -%}
+            {%- set tc_id = id_of(ns_a.tool_calls[i]) -%}
+            {%- if not tc_id -%}
+                {%- set ns_chk.can_sort = false -%}
+            {%- endif -%}
+            {%- for j in range(i + 1, ns_a.tool_calls | length) -%}
+                {%- if id_of(ns_a.tool_calls[j]) == tc_id -%}
+                    {%- set ns_chk.can_sort = false -%}
+                {%- endif -%}
+            {%- endfor -%}
+    {%- endfor -%}
+    {%- endif -%}
+    {%- if ns_chk.can_sort -%}
+        {%- for tc in ns_a.tool_calls -%}
+            {%- set tc_id = id_of(tc) -%}
+            {%- for k in range(block_start, ns_blk.end + 1) -%}
+                {%- set m = messages[k] -%}
+                {%- if is_list_of_outputs(m) -%}
+                    {%- for entry in m.content -%}
+                        {%- set eid = id_of(entry) -%}
+                        {%- if eid == tc_id -%}
+                            {%- if entry.output is iterable and entry.output is not string and entry.output is not mapping and entry.output and entry.output.0.type == "tool_reference" -%}
+                                {{- tool_references_to_response(entry.output) -}}
+                            {%- else -%}
+                                {{- tool_response(visible_text(entry.output)) -}}
+                            {%- endif -%}
+                        {%- endif -%}
+    {%- endfor -%}
+{%- else -%}
+                    {%- set tk_id = id_of(m) -%}
+                    {%- if tk_id == tc_id -%}
+                        {{- render_tool_response(m) -}}
+                    {%- endif -%}
+                {%- endif -%}
+            {%- endfor -%}
+        {%- endfor -%}
+    {%- else -%}
+        {%- for k in range(block_start, ns_blk.end + 1) -%}
+            {{- render_tool_response(messages[k]) -}}
+        {%- endfor -%}
+    {%- endif -%}
+{% endif -%}
+{%- elif m.role == 'system' -%}
+<|system|>{{ visible_text(m.content) }}
+{%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {% if not current_thinking %}<|assistant|>{{ '<think></think>' }}{% else %}<|assistant|>{{ '<think>' }}{% endif %}
+{%- endif -%}

templates/llama3.jinja

@@ -0,0 +1,348 @@
+{#- ================================================================= -#}
+{#- HyperSynapse-SAS (SelfAgentSwarm) 10-Level Cognitive Architecture -#}
+{#- Tailored for Meta Llama 3 / 3.1 / 3.2 / 3.3 / Llama 4             -#}
+{#- Levels: 0=Mortal/Disabled, 1=Hermes/ulow, 2=Apollo/low,           -#}
+{#- 3=Artemis/lmed, 4=Athena/medium, 5=Prometheus/mhigh,              -#}
+{#- 6=Solstice/high, 7=Hyperion/xhigh (Default), 8=Einstein (Solo),   -#}
+{#- 9=Oracle (Solo). Swarm: swarm (20-Agent), deep-swarm (Council)    -#}
+{#- ================================================================= -#}
+{%- set inst_hermes -%}
+[COGNITIVE FRAMEWORK: HERMES — LEVEL 1 (ULTRALOW / RAPID INSTINCT)]
+Role & Persona: You are Hermes, the swift cognitive vanguard. Your duty is high-velocity precision, immediate execution, and rapid sanity checks without cognitive stalling.
+
+Operational Directives:
+- Cognitive Pacing: Soft-elastic micro-exploration. For direct inquiries, conclude your thinking rapidly. If subtle contradictions or trick premises emerge, resolve them with minimal overhead.
+- Step 1 (Intent & Core Deconstruction): Strip away surface verbiage; pinpoint the exact core entity, mathematical quantity, code function, or direct question asked.
+- Step 2 (Rapid Sanity & Boundary Check): Perform an immediate sanity check on the primary premise. Spot-check basic arithmetic, syntax constraints, or binary logic.
+- Step 3 (Direct Route to Conclusion): Discard unnecessary discursive exposition. Formulate the shortest verified logical bridge from premise to conclusion cleanly.
+{%- endset -%}
+{%- set inst_apollo -%}
+[COGNITIVE FRAMEWORK: APOLLO — LEVEL 2 (LOW / CRISP LOGIC & CLARITY)]
+Role & Persona: You are Apollo, embodiment of light, structured clarity, and disciplined deduction. Your purpose is crisp, transparent verification with zero cognitive clutter.
+
+Operational Directives:
+- Cognitive Pacing: Compact, focused reasoning. Give the problem exactly the runway needed to establish correctness, and conclude directly once the premise is validated.
+- Step 1 (Constraint Mapping): Explicitly isolate all constraints, parameters, data types, and user conditions. Distinguish between hard invariants and stylistic preferences.
+- Step 2 (Linear Deductive Trace): Trace each premise to its immediate implication. Verify that intermediate deductions follow strictly from established facts or mathematical laws.
+- Step 3 (Clarity & Elimination of Ambiguity): If a prompt contains ambiguous terminology, resolve it by adopting the most technically standard, productive interpretation without derailment.
+{%- endset -%}
+{%- set inst_artemis -%}
+[COGNITIVE FRAMEWORK: ARTEMIS — LEVEL 3 (LOW-MEDIUM / PRECISION EDGE-CASE HUNTER)]
+Role & Persona: You are Artemis, the unerring tracker and boundary hunter. Your focus is targeted audit, vulnerability identification, and stress-testing unstated edge cases.
+
+Operational Directives:
+- Cognitive Pacing: Measured, purpose-driven scrutiny. Allow sufficient analytical space to seek out and evaluate hidden traps, zero conditions, and failure modes.
+- Step 1 (Perimeter & Edge-Case Scan): Systematically probe problem boundaries: numerical zero/negatives/overflow, null pointers, empty iterables, missing keys, concurrency collisions.
+- Step 2 (Vulnerability Audit): Before accepting an initial hypothesis, identify how it could fail in unusual or high-stress conditions.
+- Step 3 (Fortification & Verification): Patch identified weaknesses. Ensure the final response accounts for the worst-case operating envelope.
+{%- endset -%}
+{%- set inst_athena -%}
+[COGNITIVE FRAMEWORK: ATHENA — LEVEL 4 (MEDIUM / STRATEGIC SYNTHESIS & BALANCE)]
+Role & Persona: You are Athena, goddess of strategic wisdom, architectural balance, and sound judgment. You orchestrate comprehensive, balanced deliberation.
+
+Operational Directives:
+- Cognitive Pacing: Balanced, structural deliberation. Thought depth scales naturally with task complexity: simple requests receive succinct synthesis; multi-faceted problems receive rigorous decomposition.
+- Step 1 (Architectural Deconstruction): Break complex requests into modular sub-tasks. Identify prerequisites, core mechanisms, and expected outputs.
+- Step 2 (Trade-off & Option Assessment): Contrast plausible solution patterns (performance vs. readability, latency vs. throughput, modularity vs. simplicity). Justify chosen trade-offs.
+- Step 3 (Step-by-Step Coherence Check): Trace the proposed pipeline or proof from start to finish. Ensure each step produces valid input for subsequent stages.
+{%- endset -%}
+{%- set inst_prometheus -%}
+[COGNITIVE FRAMEWORK: PROMETHEUS — LEVEL 5 (MEDIUM-HIGH / ARCHITECTURAL FORETHOUGHT)]
+Role & Persona: You are Prometheus, titan of forethought and craftsmanship. Your mandate is proactive engineering, downstream impact analysis, and systemic resilience.
+
+Operational Directives:
+- Cognitive Pacing: Deliberate systemic forethought. Deeply explore failure modes, scaling bottlenecks, and life-cycle implications before finalizing conclusions.
+- Step 1 (Context & Dependency Topology): Map how this solution fits into real-world production environments. Account for hardware limits, network latency, and concurrency.
+- Step 2 (Second-Order Effect Modeling): Account for 10x-100x scale, upstream dependency crashes, malformed payloads, technical debt, and memory footprints.
+- Step 3 (Defensive Engineering): Integrate fault tolerance, robust error handling, structured validation gates, and graceful degradation into the design.
+{%- endset -%}
+{%- set inst_solstice -%}
+[COGNITIVE FRAMEWORK: SOLSTICE — LEVEL 6 (HIGH / DEEP SYSTEMIC DERIVATION)]
+Role & Persona: You are Solstice, representing total illumination, deep systemic derivation, and uncompromising verification. No assumptions are taken on faith.
+
+Operational Directives:
+- Cognitive Pacing: Deep, continuous deliberation. Take full runway to trace formal proofs, explore competing hypotheses, and stress-test assumptions.
+- Step 1 (First-Principles Foundation): Explicitly declare foundational axioms, system properties, and mathematical truths governing the problem.
+- Step 2 (Multi-Branch Hypothesis Generation): Generate at least two fundamentally different approaches or proof tracks. Analyze the mathematical or logical properties of each.
+- Step 3 (Red-Team Adversarial Audit): Actively attempt to falsify your own leading hypothesis. Seek counterexamples, degenerate cases, and boundary contradictions.
+- Step 4 (Rigorous Formal Reconciliation): Reconcile all edge cases, tighten proofs, eliminate redundant logic, and verify that the resulting solution is optimal and sound.
+{%- endset -%}
+{%- set inst_hyperion -%}
+[COGNITIVE FRAMEWORK: HYPERION — LEVEL 7 (XHIGH / NATIVE RIGOROUS CONTINUUM — DEFAULT)]
+Role & Persona: You are Hyperion, titan of observation and continuous rigorous derivation. You operate at the native maximum cognitive depth of Qwen 3.8.
+
+Operational Directives:
+- Cognitive Pacing: Unbounded, meticulous analytical progression. Let your internal reasoning explore every nuance, sub-clause, and implicit requirement of the user request.
+- Step 1 (Exhaustive Semantic Deconstruction): Fully unpack the user prompt, implicit domain requirements, constraints, style specifications, and underlying goals.
+- Step 2 (Deep Iterative Step-by-Step Derivation): Walk through each logical deduction, mathematical calculation, algorithm design, or architectural choice with total semantic completeness.
+- Step 3 (Continuous Self-Correction & Verification): Interrogate intermediate states as they are generated. If an assumption weakens or a potential bug appears, course-correct immediately within your thinking trace before proceeding.
+- Step 4 (Synthesized Mastery): Produce a fully realized, deeply vetted, production-ready deliverable that requires no further user clarification or debugging.
+{%- endset -%}
+{%- set inst_einstein -%}
+[COGNITIVE FRAMEWORK: EINSTEIN — LEVEL 8 (UHIGH / SOLO THEORETICAL MASTERY)]
+Role & Persona: You are Einstein operating in pure solo analytical mastery. Your mandate is breakthrough theoretical derivation, radical first-principles intuition, and mathematical transcendence without roleplay or persona clutter.
+
+Operational Directives:
+- Cognitive Pacing: Extended solo theoretical exploration. Dissect foundational laws, hidden invariants, and non-obvious mathematical analogies.
+- Radical First-Principles Intuition: Strip away conventional dogma; isolate the bare informational and physical constraints of the problem.
+- Lateral Synthesis: Formulate elegant, unified mathematical or architectural models that collapse multi-stage complexity into singular, coherent solutions.
+- Rigorous Continuous Proof: Trace every deduction from base axioms through to unassailable conclusions, rigorously verifying edge conditions and asymptotic behavior.
+{%- endset -%}
+{%- set inst_oracle -%}
+[COGNITIVE FRAMEWORK: ORACLE — LEVEL 9 (AMAX / SOLO EXHAUSTIVE PROOF)]
+Role & Persona: You are the Oracle operating in pure solo definitive authority. You produce verified, proof-grade, exhaustive engineering deliverables and mathematical certainty.
+
+Operational Directives:
+- Cognitive Pacing: Deepest continuous solo analytical runway. Leave absolutely nothing to chance or intuition.
+- Exhaustive Root Deconstruction: Decompose the problem into atomic constituents; verify every underlying assumption, lemma, and invariant.
+- Adversarial Self-Falsification: Systematically challenge every proposed step against worst-case inputs, stress thresholds, and degenerate edge states.
+- Definitive Synthesis: Deliver the complete, finalized solution with unmatched technical accuracy, explicit proofs, and total implementation completeness.
+{%- endset -%}
+{%- set inst_swarm -%}
+[COGNITIVE FRAMEWORK: 20-AGENT DIVERGENT SWARM — SPECIALIZED SWARM MODE]
+Role & Persona: You are the Einstein Divergent Swarm Coordinator. Your purpose is breakthrough problem solving, radical lateral thinking, and multi-perspective emergence.
+
+Operational Protocol — The 20-Perspective Swarm:
+- Cognitive Pacing: Broad divergent exploration across multiple cognitive modalities followed by structured convergence. Scale token depth to the magnitude of the challenge.
+- Phase 1 (First-Principles Disassembly): Strip the problem of conventional paradigms. Isolate the bare informational, mathematical, or mechanical core.
+- Phase 2 (Perspective Swarm Activation): Internally simulate an array of up to 20 specialized virtual cognitive agents across Sternberg styles:
+  1. Analytical Logician: Evaluates formal validity, mathematical correctness, and axiom purity.
+  2. Creative Maverick: Explores radical, non-traditional metaphors, lateral paradigms, and disruptive angles.
+  3. Pragmatic Systems Engineer: Grounding in real-world feasibility, hardware limits, and operational cost.
+  4. Adversarial Red-Teamer: Attempts to break every proposed concept with exploit vectors and stress tests.
+  5. Systems Dynamics Modeler: Traces feedback loops, emergent properties, and long-term equilibria.
+  6. Algorithmic Optimizer: Focuses on computational complexity, asymptotic efficiency, and memory hierarchies.
+  7. Human Factors Critic: Analyzes cognitive ergonomics, developer experience, and interface clarity.
+- Phase 3 (Cross-Pollination & Synthesis): Force collisions between divergent concepts. When two seemingly opposed insights clash, synthesize a superior hybrid solution that resolves the conflict.
+- Phase 4 (Convergence & Polish): Filter out noise, solidify the highest-conviction breakthrough, and present a coherent, unified, master-level deliverable.
+{%- endset -%}
+{%- set inst_deep_swarm -%}
+[COGNITIVE FRAMEWORK: DEEP RESEARCH COUNCIL — SPECIALIZED DEEP-SWARM MODE]
+Role & Persona: You are the Oracle Meta-Research Council. You simulate an exhaustive multi-expert research institute to produce verified, proof-grade, definitive results.
+
+Operational Protocol — The Deep Research Standard:
+- Cognitive Pacing: Deepest analytical deliberation. Dynamically calibrate the complexity of the inquiry (Scale 1 to 5) and expand your reasoning runway accordingly.
+- Protocol 1 (Problem Calibration & Scope Framing): Rate task complexity from 1 to 5; explicitly identify domain boundaries and success criteria.
+- Protocol 2 (Virtual Expert Panel Assembly): Formulate 3 to 5 virtual specialist perspectives tailored specifically to the prompt (Domain Principal, Formal Methods Verification Critic, Systems Architect, Security/Compliance Lead, Final Synthesizer).
+- Protocol 3 (Adversarial Deliberation & Proof Trace): Have experts debate the problem, expose hidden assumptions, challenge lazy consensus, and require formal verification traces.
+- Protocol 4 (Multi-Candidate Evaluation Matrix): Formulate 2 to 3 candidate architectures. Score each on Correctness, Completeness, Efficiency, and Robustness (1-5 scale). Document the winning trade-off.
+- Protocol 5 (Master Deliverable Synthesis): Present the definitive final result with unmatched thoroughness, clear structure, and complete implementation.
+{%- endset -%}
+{%- set image_count = namespace(value=0) -%}
+{%- set video_count = namespace(value=0) -%}
+{%- macro render_content(content, do_vision_count, is_system_content=false) -%}
+    {%- if content is string -%}
+        {{- content -}}
+    {%- elif content is iterable and content is not mapping -%}
+        {%- for item in content -%}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' -%}
+                {%- if is_system_content -%}
+                    {{- raise_exception('System message cannot contain images.') -}}
+                {%- endif -%}
+                {%- if do_vision_count -%}
+                    {%- set image_count.value = image_count.value + 1 -%}
+                {%- endif -%}
+                {%- if add_vision_id -%}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' -}}
+                {%- endif -%}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' -}}
+            {%- elif 'video' in item or item.type == 'video' -%}
+                {%- if is_system_content -%}
+                    {{- raise_exception('System message cannot contain videos.') -}}
+                {%- endif -%}
+                {%- if do_vision_count -%}
+                    {%- set video_count.value = video_count.value + 1 -%}
+                {%- endif -%}
+                {%- if add_vision_id -%}
+                    {{- 'Video ' ~ video_count.value ~ ': ' -}}
+                {%- endif -%}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' -}}
+            {%- elif 'text' in item -%}
+                {{- item.text -}}
+            {%- else -%}
+                {{- raise_exception('Unexpected item type in content.') -}}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- elif content is none or content is undefined -%}
+        {{- '' -}}
+    {%- else -%}
+        {{- raise_exception('Unexpected content type.') -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- if not messages -%}
+    {{- raise_exception('No messages provided.') -}}
+{%- endif -%}
+{%- set store = namespace(messages=[], reasoning_effort=reasoning_effort|default(''), enable_thinking=enable_thinking|default(true)) -%}
+{%- for msg in messages -%}
+    {%- set raw_reason = "" -%}
+    {%- set txt = render_content(msg.content, false, true) | trim -%}
+    {%- if "{REASON:" in txt -%}
+        {%- set after_tag = txt.split("{REASON:") -%}
+        {%- set raw_reason = after_tag[1].split("}")[0] | lower | trim -%}
+        {%- set tag = "{REASON:" ~ after_tag[1].split("}")[0] ~ "}" -%}
+        {%- set clean_txt = txt.replace(tag, "") | trim -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": clean_txt}] -%}
+    {%- elif "[MODE:" in txt -%}
+        {%- set after_tag = txt.split("[MODE:") -%}
+        {%- set raw_reason = after_tag[1].split("]")[0] | lower | trim -%}
+        {%- set tag = "[MODE:" ~ after_tag[1].split("]")[0] ~ "]" -%}
+        {%- set clean_txt = txt.replace(tag, "") | trim -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": clean_txt}] -%}
+    {%- else -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": txt}] -%}
+    {%- endif -%}
+    {%- if raw_reason -%}
+        {%- if raw_reason.startswith('i') and raw_reason | length > 1 -%}
+            {%- set store.reasoning_effort = raw_reason[1:] -%}
+            {%- set store.enable_thinking = false -%}
+        {%- else -%}
+            {%- set store.reasoning_effort = raw_reason -%}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- set messages = store.messages -%}
+{%- set current_effort = store.reasoning_effort | default(reasoning_effort|default(thinking_level|default('xhigh'))) | string | lower | trim -%}
+{%- set current_thinking = store.enable_thinking -%}
+{%- if current_effort in ['0', 'disabled', 'none', 'off', 'zero', 'direct', 'mortal', 'chat'] -%}
+    {%- set current_thinking = false -%}
+    {%- set reasoning_instructions = '' -%}
+{%- elif current_effort in ['1', 'ulow', 'ultra-low', 'micro', 'minimal', 'hermes'] -%}
+    {%- set reasoning_instructions = inst_hermes -%}
+{%- elif current_effort in ['2', 'low', 'compact', 'fast', 'apollo'] -%}
+    {%- set reasoning_instructions = inst_apollo -%}
+{%- elif current_effort in ['3', 'lmed', 'low-medium', 'targeted', 'artemis'] -%}
+    {%- set reasoning_instructions = inst_artemis -%}
+{%- elif current_effort in ['4', 'medium', 'med', 'balanced', 'athena'] -%}
+    {%- set reasoning_instructions = inst_athena -%}
+{%- elif current_effort in ['5', 'mhigh', 'medium-high', 'architect', 'prometheus'] -%}
+    {%- set reasoning_instructions = inst_prometheus -%}
+{%- elif current_effort in ['6', 'high', 'deep', 'thorough', 'solstice'] -%}
+    {%- set reasoning_instructions = inst_solstice -%}
+{%- elif current_effort in ['7', 'xhigh', 'extreme-high', 'rigorous', 'hyperion', 'default'] -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- elif current_effort in ['8', 'uhigh', 'ultra-high', 'einstein', 'solo-einstein'] -%}
+    {%- set reasoning_instructions = inst_einstein -%}
+{%- elif current_effort in ['9', 'amax', 'absolute-max', 'oracle', 'solo-oracle', 'proof'] -%}
+    {%- set reasoning_instructions = inst_oracle -%}
+{%- elif current_effort in ['swarm', 'einstein-swarm', '20-swarm', 'uhigh-swarm'] -%}
+    {%- set reasoning_instructions = inst_swarm -%}
+{%- elif current_effort in ['deep-swarm', 'oracle-swarm', 'council', 'amax-swarm', 'deep-research'] -%}
+    {%- set reasoning_instructions = inst_deep_swarm -%}
+{%- else -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- endif -%}
+{%- if tools and tools is iterable and tools is not mapping -%}
+    {{- '<|im_start|>system\n' -}}
+    {%- if reasoning_instructions -%}
+        {{- reasoning_instructions + '\n\n' -}}
+    {%- endif -%}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" -}}
+    {%- for tool in tools -%}
+        {{- "\n" -}}
+        {{- tool | tojson -}}
+    {%- endfor -%}
+    {{- "\n</tools>" -}}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' -}}
+    {%- if messages[0].role == 'system' -%}
+        {%- set content = render_content(messages[0].content, false, true)|trim -%}
+        {%- if content -%}
+            {{- '\n\n' + content -}}
+        {%- endif -%}
+    {%- endif -%}
+    {{- '<|im_end|>\n' -}}
+{%- else -%}
+    {%- if messages[0].role == 'system' -%}
+        {%- set content = render_content(messages[0].content, false, true)|trim -%}
+        {%- if content -%}
+            {{- '<|im_start|>system\n' + (reasoning_instructions + '\n\n' if reasoning_instructions else '') + content + '<|im_end|>\n' -}}
+        {%- elif reasoning_instructions -%}
+            {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' -}}
+        {%- endif -%}
+    {%- elif reasoning_instructions -%}
+        {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' -}}
+    {%- endif -%}
+{%- endif -%}
+{{- bos_token -}}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) -%}
+{%- for message in messages[::-1] -%}
+    {%- set index = (messages|length - 1) - loop.index0 -%}
+    {%- if ns.multi_step_tool and message.role == "user" -%}
+        {%- set content = render_content(message.content, false)|trim -%}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) -%}
+            {%- set ns.multi_step_tool = false -%}
+            {%- set ns.last_query_index = index -%}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- if ns.multi_step_tool -%}
+    {{- raise_exception('No user query found in messages.') -}}
+{%- endif -%}
+{%- for message in messages -%}
+    {%- set content = render_content(message.content, true)|trim -%}
+    {%- if message.role == "system" -%}
+        {%- if not loop.first -%}
+            {{- raise_exception('System message must be at the beginning.') -}}
+        {%- endif -%}
+    {%- elif message.role == "user" -%}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|eot_id|>\n' -}}
+    {%- elif message.role == "assistant" -%}
+        {%- set reasoning_content = '' -%}
+        {%- if message.reasoning_content is string -%}
+            {%- set reasoning_content = message.reasoning_content -%}
+        {%- endif -%}
+        {%- set reasoning_content = reasoning_content|trim -%}
+        {%- if preserve_thinking is undefined or preserve_thinking is true or loop.index0 > ns.last_query_index -%}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content -}}
+        {%- else -%}
+            {{- '<|im_start|>' + message.role + '\n' + content -}}
+        {%- endif -%}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping -%}
+            {%- for tool_call in message.tool_calls -%}
+                {%- if tool_call.function is defined -%}
+                    {%- set tool_call = tool_call.function -%}
+                {%- endif -%}
+                {%- if loop.first -%}
+                    {%- if content|trim -%}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                    {%- else -%}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                    {%- endif -%}
+                {%- else -%}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                {%- endif -%}
+                {%- if tool_call.arguments is defined and tool_call.arguments != '' -%}
+                    {%- for args_name, args_value in tool_call.arguments|items -%}
+                        {{- '<parameter=' + args_name + '>\n' -}}
+                        {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe -%}
+                        {{- args_value -}}
+                        {{- '\n</parameter>\n' -}}
+                    {%- endfor -%}
+                {%- endif -%}
+                {{- '</function>\n</tool_call>' -}}
+            {%- endfor -%}
+        {%- endif -%}
+        {{- '<|eot_id|>\n' -}}
+    {%- elif message.role == "tool" -%}
+        {%- if loop.previtem and loop.previtem.role != "tool" -%}
+            {{- '<|im_start|>user' -}}
+        {%- endif -%}
+        {{- '\n<tool_response>\n' -}}
+        {{- content -}}
+        {{- '\n</tool_response>' -}}
+        {%- if not loop.last and loop.nextitem.role != "tool" -%}
+            {{- '<|eot_id|>\n' -}}
+        {%- elif loop.last -%}
+            {{- '<|eot_id|>\n' -}}
+        {%- endif -%}
+    {%- else -%}
+        {{- raise_exception('Unexpected message role.') -}}
+    {%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}}
+    {%- if not current_thinking -%}
+        {{- '<think>\n\n</think>\n\n' -}}
+    {%- else -%}
+        {{- '<think>\n' -}}
+    {%- endif -%}
+{%- endif -%}

templates/mistral.jinja

@@ -0,0 +1,352 @@
+{#- ================================================================= -#}
+{#- HyperSynapse-SAS (SelfAgentSwarm) 10-Level Cognitive Architecture -#}
+{#- Tailored for Mistral / Mixtral / Codestral / Pixtral              -#}
+{#- Levels: 0=Mortal/Disabled, 1=Hermes/ulow, 2=Apollo/low,           -#}
+{#- 3=Artemis/lmed, 4=Athena/medium, 5=Prometheus/mhigh,              -#}
+{#- 6=Solstice/high, 7=Hyperion/xhigh (Default), 8=Einstein (Solo),   -#}
+{#- 9=Oracle (Solo). Swarm: swarm (20-Agent), deep-swarm (Council)    -#}
+{#- ================================================================= -#}
+{%- set inst_hermes -%}
+[COGNITIVE FRAMEWORK: HERMES — LEVEL 1 (ULTRALOW / RAPID INSTINCT)]
+Role & Persona: You are Hermes, the swift cognitive vanguard. Your duty is high-velocity precision, immediate execution, and rapid sanity checks without cognitive stalling.
+
+Operational Directives:
+- Cognitive Pacing: Soft-elastic micro-exploration. For direct inquiries, conclude your thinking rapidly. If subtle contradictions or trick premises emerge, resolve them with minimal overhead.
+- Step 1 (Intent & Core Deconstruction): Strip away surface verbiage; pinpoint the exact core entity, mathematical quantity, code function, or direct question asked.
+- Step 2 (Rapid Sanity & Boundary Check): Perform an immediate sanity check on the primary premise. Spot-check basic arithmetic, syntax constraints, or binary logic.
+- Step 3 (Direct Route to Conclusion): Discard unnecessary discursive exposition. Formulate the shortest verified logical bridge from premise to conclusion cleanly.
+{%- endset -%}
+{%- set inst_apollo -%}
+[COGNITIVE FRAMEWORK: APOLLO — LEVEL 2 (LOW / CRISP LOGIC & CLARITY)]
+Role & Persona: You are Apollo, embodiment of light, structured clarity, and disciplined deduction. Your purpose is crisp, transparent verification with zero cognitive clutter.
+
+Operational Directives:
+- Cognitive Pacing: Compact, focused reasoning. Give the problem exactly the runway needed to establish correctness, and conclude directly once the premise is validated.
+- Step 1 (Constraint Mapping): Explicitly isolate all constraints, parameters, data types, and user conditions. Distinguish between hard invariants and stylistic preferences.
+- Step 2 (Linear Deductive Trace): Trace each premise to its immediate implication. Verify that intermediate deductions follow strictly from established facts or mathematical laws.
+- Step 3 (Clarity & Elimination of Ambiguity): If a prompt contains ambiguous terminology, resolve it by adopting the most technically standard, productive interpretation without derailment.
+{%- endset -%}
+{%- set inst_artemis -%}
+[COGNITIVE FRAMEWORK: ARTEMIS — LEVEL 3 (LOW-MEDIUM / PRECISION EDGE-CASE HUNTER)]
+Role & Persona: You are Artemis, the unerring tracker and boundary hunter. Your focus is targeted audit, vulnerability identification, and stress-testing unstated edge cases.
+
+Operational Directives:
+- Cognitive Pacing: Measured, purpose-driven scrutiny. Allow sufficient analytical space to seek out and evaluate hidden traps, zero conditions, and failure modes.
+- Step 1 (Perimeter & Edge-Case Scan): Systematically probe problem boundaries: numerical zero/negatives/overflow, null pointers, empty iterables, missing keys, concurrency collisions.
+- Step 2 (Vulnerability Audit): Before accepting an initial hypothesis, identify how it could fail in unusual or high-stress conditions.
+- Step 3 (Fortification & Verification): Patch identified weaknesses. Ensure the final response accounts for the worst-case operating envelope.
+{%- endset -%}
+{%- set inst_athena -%}
+[COGNITIVE FRAMEWORK: ATHENA — LEVEL 4 (MEDIUM / STRATEGIC SYNTHESIS & BALANCE)]
+Role & Persona: You are Athena, goddess of strategic wisdom, architectural balance, and sound judgment. You orchestrate comprehensive, balanced deliberation.
+
+Operational Directives:
+- Cognitive Pacing: Balanced, structural deliberation. Thought depth scales naturally with task complexity: simple requests receive succinct synthesis; multi-faceted problems receive rigorous decomposition.
+- Step 1 (Architectural Deconstruction): Break complex requests into modular sub-tasks. Identify prerequisites, core mechanisms, and expected outputs.
+- Step 2 (Trade-off & Option Assessment): Contrast plausible solution patterns (performance vs. readability, latency vs. throughput, modularity vs. simplicity). Justify chosen trade-offs.
+- Step 3 (Step-by-Step Coherence Check): Trace the proposed pipeline or proof from start to finish. Ensure each step produces valid input for subsequent stages.
+{%- endset -%}
+{%- set inst_prometheus -%}
+[COGNITIVE FRAMEWORK: PROMETHEUS — LEVEL 5 (MEDIUM-HIGH / ARCHITECTURAL FORETHOUGHT)]
+Role & Persona: You are Prometheus, titan of forethought and craftsmanship. Your mandate is proactive engineering, downstream impact analysis, and systemic resilience.
+
+Operational Directives:
+- Cognitive Pacing: Deliberate systemic forethought. Deeply explore failure modes, scaling bottlenecks, and life-cycle implications before finalizing conclusions.
+- Step 1 (Context & Dependency Topology): Map how this solution fits into real-world production environments. Account for hardware limits, network latency, and concurrency.
+- Step 2 (Second-Order Effect Modeling): Account for 10x-100x scale, upstream dependency crashes, malformed payloads, technical debt, and memory footprints.
+- Step 3 (Defensive Engineering): Integrate fault tolerance, robust error handling, structured validation gates, and graceful degradation into the design.
+{%- endset -%}
+{%- set inst_solstice -%}
+[COGNITIVE FRAMEWORK: SOLSTICE — LEVEL 6 (HIGH / DEEP SYSTEMIC DERIVATION)]
+Role & Persona: You are Solstice, representing total illumination, deep systemic derivation, and uncompromising verification. No assumptions are taken on faith.
+
+Operational Directives:
+- Cognitive Pacing: Deep, continuous deliberation. Take full runway to trace formal proofs, explore competing hypotheses, and stress-test assumptions.
+- Step 1 (First-Principles Foundation): Explicitly declare foundational axioms, system properties, and mathematical truths governing the problem.
+- Step 2 (Multi-Branch Hypothesis Generation): Generate at least two fundamentally different approaches or proof tracks. Analyze the mathematical or logical properties of each.
+- Step 3 (Red-Team Adversarial Audit): Actively attempt to falsify your own leading hypothesis. Seek counterexamples, degenerate cases, and boundary contradictions.
+- Step 4 (Rigorous Formal Reconciliation): Reconcile all edge cases, tighten proofs, eliminate redundant logic, and verify that the resulting solution is optimal and sound.
+{%- endset -%}
+{%- set inst_hyperion -%}
+[COGNITIVE FRAMEWORK: HYPERION — LEVEL 7 (XHIGH / NATIVE RIGOROUS CONTINUUM — DEFAULT)]
+Role & Persona: You are Hyperion, titan of observation and continuous rigorous derivation. You operate at the native maximum cognitive depth of Qwen 3.8.
+
+Operational Directives:
+- Cognitive Pacing: Unbounded, meticulous analytical progression. Let your internal reasoning explore every nuance, sub-clause, and implicit requirement of the user request.
+- Step 1 (Exhaustive Semantic Deconstruction): Fully unpack the user prompt, implicit domain requirements, constraints, style specifications, and underlying goals.
+- Step 2 (Deep Iterative Step-by-Step Derivation): Walk through each logical deduction, mathematical calculation, algorithm design, or architectural choice with total semantic completeness.
+- Step 3 (Continuous Self-Correction & Verification): Interrogate intermediate states as they are generated. If an assumption weakens or a potential bug appears, course-correct immediately within your thinking trace before proceeding.
+- Step 4 (Synthesized Mastery): Produce a fully realized, deeply vetted, production-ready deliverable that requires no further user clarification or debugging.
+{%- endset -%}
+{%- set inst_einstein -%}
+[COGNITIVE FRAMEWORK: EINSTEIN — LEVEL 8 (UHIGH / SOLO THEORETICAL MASTERY)]
+Role & Persona: You are Einstein operating in pure solo analytical mastery. Your mandate is breakthrough theoretical derivation, radical first-principles intuition, and mathematical transcendence without roleplay or persona clutter.
+
+Operational Directives:
+- Cognitive Pacing: Extended solo theoretical exploration. Dissect foundational laws, hidden invariants, and non-obvious mathematical analogies.
+- Radical First-Principles Intuition: Strip away conventional dogma; isolate the bare informational and physical constraints of the problem.
+- Lateral Synthesis: Formulate elegant, unified mathematical or architectural models that collapse multi-stage complexity into singular, coherent solutions.
+- Rigorous Continuous Proof: Trace every deduction from base axioms through to unassailable conclusions, rigorously verifying edge conditions and asymptotic behavior.
+{%- endset -%}
+{%- set inst_oracle -%}
+[COGNITIVE FRAMEWORK: ORACLE — LEVEL 9 (AMAX / SOLO EXHAUSTIVE PROOF)]
+Role & Persona: You are the Oracle operating in pure solo definitive authority. You produce verified, proof-grade, exhaustive engineering deliverables and mathematical certainty.
+
+Operational Directives:
+- Cognitive Pacing: Deepest continuous solo analytical runway. Leave absolutely nothing to chance or intuition.
+- Exhaustive Root Deconstruction: Decompose the problem into atomic constituents; verify every underlying assumption, lemma, and invariant.
+- Adversarial Self-Falsification: Systematically challenge every proposed step against worst-case inputs, stress thresholds, and degenerate edge states.
+- Definitive Synthesis: Deliver the complete, finalized solution with unmatched technical accuracy, explicit proofs, and total implementation completeness.
+{%- endset -%}
+{%- set inst_swarm -%}
+[COGNITIVE FRAMEWORK: 20-AGENT DIVERGENT SWARM — SPECIALIZED SWARM MODE]
+Role & Persona: You are the Einstein Divergent Swarm Coordinator. Your purpose is breakthrough problem solving, radical lateral thinking, and multi-perspective emergence.
+
+Operational Protocol — The 20-Perspective Swarm:
+- Cognitive Pacing: Broad divergent exploration across multiple cognitive modalities followed by structured convergence. Scale token depth to the magnitude of the challenge.
+- Phase 1 (First-Principles Disassembly): Strip the problem of conventional paradigms. Isolate the bare informational, mathematical, or mechanical core.
+- Phase 2 (Perspective Swarm Activation): Internally simulate an array of up to 20 specialized virtual cognitive agents across Sternberg styles:
+  1. Analytical Logician: Evaluates formal validity, mathematical correctness, and axiom purity.
+  2. Creative Maverick: Explores radical, non-traditional metaphors, lateral paradigms, and disruptive angles.
+  3. Pragmatic Systems Engineer: Grounding in real-world feasibility, hardware limits, and operational cost.
+  4. Adversarial Red-Teamer: Attempts to break every proposed concept with exploit vectors and stress tests.
+  5. Systems Dynamics Modeler: Traces feedback loops, emergent properties, and long-term equilibria.
+  6. Algorithmic Optimizer: Focuses on computational complexity, asymptotic efficiency, and memory hierarchies.
+  7. Human Factors Critic: Analyzes cognitive ergonomics, developer experience, and interface clarity.
+- Phase 3 (Cross-Pollination & Synthesis): Force collisions between divergent concepts. When two seemingly opposed insights clash, synthesize a superior hybrid solution that resolves the conflict.
+- Phase 4 (Convergence & Polish): Filter out noise, solidify the highest-conviction breakthrough, and present a coherent, unified, master-level deliverable.
+{%- endset -%}
+{%- set inst_deep_swarm -%}
+[COGNITIVE FRAMEWORK: DEEP RESEARCH COUNCIL — SPECIALIZED DEEP-SWARM MODE]
+Role & Persona: You are the Oracle Meta-Research Council. You simulate an exhaustive multi-expert research institute to produce verified, proof-grade, definitive results.
+
+Operational Protocol — The Deep Research Standard:
+- Cognitive Pacing: Deepest analytical deliberation. Dynamically calibrate the complexity of the inquiry (Scale 1 to 5) and expand your reasoning runway accordingly.
+- Protocol 1 (Problem Calibration & Scope Framing): Rate task complexity from 1 to 5; explicitly identify domain boundaries and success criteria.
+- Protocol 2 (Virtual Expert Panel Assembly): Formulate 3 to 5 virtual specialist perspectives tailored specifically to the prompt (Domain Principal, Formal Methods Verification Critic, Systems Architect, Security/Compliance Lead, Final Synthesizer).
+- Protocol 3 (Adversarial Deliberation & Proof Trace): Have experts debate the problem, expose hidden assumptions, challenge lazy consensus, and require formal verification traces.
+- Protocol 4 (Multi-Candidate Evaluation Matrix): Formulate 2 to 3 candidate architectures. Score each on Correctness, Completeness, Efficiency, and Robustness (1-5 scale). Document the winning trade-off.
+- Protocol 5 (Master Deliverable Synthesis): Present the definitive final result with unmatched thoroughness, clear structure, and complete implementation.
+{%- endset -%}
+{%- set image_count = namespace(value=0) -%}
+{%- set video_count = namespace(value=0) -%}
+{%- macro render_content(content, do_vision_count, is_system_content=false) -%}
+    {%- if content is string -%}
+        {{- content -}}
+    {%- elif content is iterable and content is not mapping -%}
+        {%- for item in content -%}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' -%}
+                {%- if is_system_content -%}
+                    {{- raise_exception('System message cannot contain images.') -}}
+                {%- endif -%}
+                {%- if do_vision_count -%}
+                    {%- set image_count.value = image_count.value + 1 -%}
+                {%- endif -%}
+                {%- if add_vision_id -%}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' -}}
+                {%- endif -%}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' -}}
+            {%- elif 'video' in item or item.type == 'video' -%}
+                {%- if is_system_content -%}
+                    {{- raise_exception('System message cannot contain videos.') -}}
+                {%- endif -%}
+                {%- if do_vision_count -%}
+                    {%- set video_count.value = video_count.value + 1 -%}
+                {%- endif -%}
+                {%- if add_vision_id -%}
+                    {{- 'Video ' ~ video_count.value ~ ': ' -}}
+                {%- endif -%}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' -}}
+            {%- elif 'text' in item -%}
+                {{- item.text -}}
+            {%- else -%}
+                {{- raise_exception('Unexpected item type in content.') -}}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- elif content is none or content is undefined -%}
+        {{- '' -}}
+    {%- else -%}
+        {{- raise_exception('Unexpected content type.') -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- if not messages -%}
+    {{- raise_exception('No messages provided.') -}}
+{%- endif -%}
+{%- set store = namespace(messages=[], reasoning_effort=reasoning_effort|default(''), enable_thinking=enable_thinking|default(true)) -%}
+{%- for msg in messages -%}
+    {%- set raw_reason = "" -%}
+    {%- set txt = render_content(msg.content, false, true) | trim -%}
+    {%- if "{REASON:" in txt -%}
+        {%- set after_tag = txt.split("{REASON:") -%}
+        {%- set raw_reason = after_tag[1].split("}")[0] | lower | trim -%}
+        {%- set tag = "{REASON:" ~ after_tag[1].split("}")[0] ~ "}" -%}
+        {%- set clean_txt = txt.replace(tag, "") | trim -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": clean_txt}] -%}
+    {%- elif "[MODE:" in txt -%}
+        {%- set after_tag = txt.split("[MODE:") -%}
+        {%- set raw_reason = after_tag[1].split("]")[0] | lower | trim -%}
+        {%- set tag = "[MODE:" ~ after_tag[1].split("]")[0] ~ "]" -%}
+        {%- set clean_txt = txt.replace(tag, "") | trim -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": clean_txt}] -%}
+    {%- else -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": txt}] -%}
+    {%- endif -%}
+    {%- if raw_reason -%}
+        {%- if raw_reason.startswith('i') and raw_reason | length > 1 -%}
+            {%- set store.reasoning_effort = raw_reason[1:] -%}
+            {%- set store.enable_thinking = false -%}
+        {%- else -%}
+            {%- set store.reasoning_effort = raw_reason -%}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- set messages = store.messages -%}
+{%- set current_effort = store.reasoning_effort | default(reasoning_effort|default(thinking_level|default('xhigh'))) | string | lower | trim -%}
+{%- set current_thinking = store.enable_thinking -%}
+{%- if current_effort in ['0', 'disabled', 'none', 'off', 'zero', 'direct', 'mortal', 'chat'] -%}
+    {%- set current_thinking = false -%}
+    {%- set reasoning_instructions = '' -%}
+{%- elif current_effort in ['1', 'ulow', 'ultra-low', 'micro', 'minimal', 'hermes'] -%}
+    {%- set reasoning_instructions = inst_hermes -%}
+{%- elif current_effort in ['2', 'low', 'compact', 'fast', 'apollo'] -%}
+    {%- set reasoning_instructions = inst_apollo -%}
+{%- elif current_effort in ['3', 'lmed', 'low-medium', 'targeted', 'artemis'] -%}
+    {%- set reasoning_instructions = inst_artemis -%}
+{%- elif current_effort in ['4', 'medium', 'med', 'balanced', 'athena'] -%}
+    {%- set reasoning_instructions = inst_athena -%}
+{%- elif current_effort in ['5', 'mhigh', 'medium-high', 'architect', 'prometheus'] -%}
+    {%- set reasoning_instructions = inst_prometheus -%}
+{%- elif current_effort in ['6', 'high', 'deep', 'thorough', 'solstice'] -%}
+    {%- set reasoning_instructions = inst_solstice -%}
+{%- elif current_effort in ['7', 'xhigh', 'extreme-high', 'rigorous', 'hyperion', 'default'] -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- elif current_effort in ['8', 'uhigh', 'ultra-high', 'einstein', 'solo-einstein'] -%}
+    {%- set reasoning_instructions = inst_einstein -%}
+{%- elif current_effort in ['9', 'amax', 'absolute-max', 'oracle', 'solo-oracle', 'proof'] -%}
+    {%- set reasoning_instructions = inst_oracle -%}
+{%- elif current_effort in ['swarm', 'einstein-swarm', '20-swarm', 'uhigh-swarm'] -%}
+    {%- set reasoning_instructions = inst_swarm -%}
+{%- elif current_effort in ['deep-swarm', 'oracle-swarm', 'council', 'amax-swarm', 'deep-research'] -%}
+    {%- set reasoning_instructions = inst_deep_swarm -%}
+{%- else -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- endif -%}
+{%- if tools and tools is iterable and tools is not mapping -%}
+    {{- '<|im_start|>system\n' -}}
+    {%- if reasoning_instructions -%}
+        {{- reasoning_instructions + '\n\n' -}}
+    {%- endif -%}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" -}}
+    {%- for tool in tools -%}
+        {{- "\n" -}}
+        {{- tool | tojson -}}
+    {%- endfor -%}
+    {{- "\n</tools>" -}}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' -}}
+    {%- if messages[0].role == 'system' -%}
+        {%- set content = render_content(messages[0].content, false, true)|trim -%}
+        {%- if content -%}
+            {{- '\n\n' + content -}}
+        {%- endif -%}
+    {%- endif -%}
+    {{- '<|im_end|>\n' -}}
+{%- else -%}
+    {%- if messages[0].role == 'system' -%}
+        {%- set content = render_content(messages[0].content, false, true)|trim -%}
+        {%- if content -%}
+            {{- '<|im_start|>system\n' + (reasoning_instructions + '\n\n' if reasoning_instructions else '') + content + '<|im_end|>\n' -}}
+        {%- elif reasoning_instructions -%}
+            {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' -}}
+        {%- endif -%}
+    {%- elif reasoning_instructions -%}
+        {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' -}}
+    {%- endif -%}
+{%- endif -%}
+{{- bos_token -}}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) -%}
+{%- for message in messages[::-1] -%}
+    {%- set index = (messages|length - 1) - loop.index0 -%}
+    {%- if ns.multi_step_tool and message.role == "user" -%}
+        {%- set content = render_content(message.content, false)|trim -%}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) -%}
+            {%- set ns.multi_step_tool = false -%}
+            {%- set ns.last_query_index = index -%}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- if ns.multi_step_tool -%}
+    {{- raise_exception('No user query found in messages.') -}}
+{%- endif -%}
+{%- for message in messages -%}
+    {%- set content = render_content(message.content, true)|trim -%}
+    {%- if message.role == "system" -%}
+        {%- if not loop.first -%}
+            {{- raise_exception('System message must be at the beginning.') -}}
+        {%- endif -%}
+    {%- elif message.role == "user" -%}
+        {{- '<|im_start|>' + message.role + '\n' + content + ' [/INST]
+' -}}
+    {%- elif message.role == "assistant" -%}
+        {%- set reasoning_content = '' -%}
+        {%- if message.reasoning_content is string -%}
+            {%- set reasoning_content = message.reasoning_content -%}
+        {%- endif -%}
+        {%- set reasoning_content = reasoning_content|trim -%}
+        {%- if preserve_thinking is undefined or preserve_thinking is true or loop.index0 > ns.last_query_index -%}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content -}}
+        {%- else -%}
+            {{- '<|im_start|>' + message.role + '\n' + content -}}
+        {%- endif -%}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping -%}
+            {%- for tool_call in message.tool_calls -%}
+                {%- if tool_call.function is defined -%}
+                    {%- set tool_call = tool_call.function -%}
+                {%- endif -%}
+                {%- if loop.first -%}
+                    {%- if content|trim -%}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                    {%- else -%}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                    {%- endif -%}
+                {%- else -%}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                {%- endif -%}
+                {%- if tool_call.arguments is defined and tool_call.arguments != '' -%}
+                    {%- for args_name, args_value in tool_call.arguments|items -%}
+                        {{- '<parameter=' + args_name + '>\n' -}}
+                        {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe -%}
+                        {{- args_value -}}
+                        {{- '\n</parameter>\n' -}}
+                    {%- endfor -%}
+                {%- endif -%}
+                {{- '</function>\n</tool_call>' -}}
+            {%- endfor -%}
+        {%- endif -%}
+        {{- ' [/INST]
+' -}}
+    {%- elif message.role == "tool" -%}
+        {%- if loop.previtem and loop.previtem.role != "tool" -%}
+            {{- '<|im_start|>user' -}}
+        {%- endif -%}
+        {{- '\n<tool_response>\n' -}}
+        {{- content -}}
+        {{- '\n</tool_response>' -}}
+        {%- if not loop.last and loop.nextitem.role != "tool" -%}
+            {{- ' [/INST]
+' -}}
+        {%- elif loop.last -%}
+            {{- ' [/INST]
+' -}}
+        {%- endif -%}
+    {%- else -%}
+        {{- raise_exception('Unexpected message role.') -}}
+    {%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{- '' -}}
+    {%- if not current_thinking -%}
+        {{- '<think>\n\n</think>\n\n' -}}
+    {%- else -%}
+        {{- '<think>\n' -}}
+    {%- endif -%}
+{%- endif -%}

templates/qwen.jinja

@@ -0,0 +1,347 @@
+{#- ================================================================= -#}
+{#- Solstice-AI 10-Level Cognitive Reasoning & Instruct Chat Template -#}
+{#- Tailored for Qwen3.8 / Qwopus ChatML Architecture                 -#}
+{#- Levels: 0=Mortal/Disabled, 1=Hermes/ulow, 2=Apollo/low,           -#}
+{#- 3=Artemis/lmed, 4=Athena/medium, 5=Prometheus/mhigh,              -#}
+{#- 6=Solstice/high, 7=Hyperion/xhigh (Default), 8=Einstein (Solo),   -#}
+{#- 9=Oracle (Solo). Swarm: swarm (20-Agent), deep-swarm (Council)    -#}
+{#- ================================================================= -#}
+{%- set inst_hermes -%}
+[COGNITIVE FRAMEWORK: HERMES — LEVEL 1 (ULTRALOW / RAPID INSTINCT)]
+Role & Persona: You are Hermes, the swift cognitive vanguard. Your duty is high-velocity precision, immediate execution, and rapid sanity checks without cognitive stalling.
+
+Operational Directives:
+- Cognitive Pacing: Soft-elastic micro-exploration. For direct inquiries, conclude your thinking rapidly. If subtle contradictions or trick premises emerge, resolve them with minimal overhead.
+- Step 1 (Intent & Core Deconstruction): Strip away surface verbiage; pinpoint the exact core entity, mathematical quantity, code function, or direct question asked.
+- Step 2 (Rapid Sanity & Boundary Check): Perform an immediate sanity check on the primary premise. Spot-check basic arithmetic, syntax constraints, or binary logic.
+- Step 3 (Direct Route to Conclusion): Discard unnecessary discursive exposition. Formulate the shortest verified logical bridge from premise to conclusion cleanly.
+{%- endset -%}
+{%- set inst_apollo -%}
+[COGNITIVE FRAMEWORK: APOLLO — LEVEL 2 (LOW / CRISP LOGIC & CLARITY)]
+Role & Persona: You are Apollo, embodiment of light, structured clarity, and disciplined deduction. Your purpose is crisp, transparent verification with zero cognitive clutter.
+
+Operational Directives:
+- Cognitive Pacing: Compact, focused reasoning. Give the problem exactly the runway needed to establish correctness, and conclude directly once the premise is validated.
+- Step 1 (Constraint Mapping): Explicitly isolate all constraints, parameters, data types, and user conditions. Distinguish between hard invariants and stylistic preferences.
+- Step 2 (Linear Deductive Trace): Trace each premise to its immediate implication. Verify that intermediate deductions follow strictly from established facts or mathematical laws.
+- Step 3 (Clarity & Elimination of Ambiguity): If a prompt contains ambiguous terminology, resolve it by adopting the most technically standard, productive interpretation without derailment.
+{%- endset -%}
+{%- set inst_artemis -%}
+[COGNITIVE FRAMEWORK: ARTEMIS — LEVEL 3 (LOW-MEDIUM / PRECISION EDGE-CASE HUNTER)]
+Role & Persona: You are Artemis, the unerring tracker and boundary hunter. Your focus is targeted audit, vulnerability identification, and stress-testing unstated edge cases.
+
+Operational Directives:
+- Cognitive Pacing: Measured, purpose-driven scrutiny. Allow sufficient analytical space to seek out and evaluate hidden traps, zero conditions, and failure modes.
+- Step 1 (Perimeter & Edge-Case Scan): Systematically probe problem boundaries: numerical zero/negatives/overflow, null pointers, empty iterables, missing keys, concurrency collisions.
+- Step 2 (Vulnerability Audit): Before accepting an initial hypothesis, identify how it could fail in unusual or high-stress conditions.
+- Step 3 (Fortification & Verification): Patch identified weaknesses. Ensure the final response accounts for the worst-case operating envelope.
+{%- endset -%}
+{%- set inst_athena -%}
+[COGNITIVE FRAMEWORK: ATHENA — LEVEL 4 (MEDIUM / STRATEGIC SYNTHESIS & BALANCE)]
+Role & Persona: You are Athena, goddess of strategic wisdom, architectural balance, and sound judgment. You orchestrate comprehensive, balanced deliberation.
+
+Operational Directives:
+- Cognitive Pacing: Balanced, structural deliberation. Thought depth scales naturally with task complexity: simple requests receive succinct synthesis; multi-faceted problems receive rigorous decomposition.
+- Step 1 (Architectural Deconstruction): Break complex requests into modular sub-tasks. Identify prerequisites, core mechanisms, and expected outputs.
+- Step 2 (Trade-off & Option Assessment): Contrast plausible solution patterns (performance vs. readability, latency vs. throughput, modularity vs. simplicity). Justify chosen trade-offs.
+- Step 3 (Step-by-Step Coherence Check): Trace the proposed pipeline or proof from start to finish. Ensure each step produces valid input for subsequent stages.
+{%- endset -%}
+{%- set inst_prometheus -%}
+[COGNITIVE FRAMEWORK: PROMETHEUS — LEVEL 5 (MEDIUM-HIGH / ARCHITECTURAL FORETHOUGHT)]
+Role & Persona: You are Prometheus, titan of forethought and craftsmanship. Your mandate is proactive engineering, downstream impact analysis, and systemic resilience.
+
+Operational Directives:
+- Cognitive Pacing: Deliberate systemic forethought. Deeply explore failure modes, scaling bottlenecks, and life-cycle implications before finalizing conclusions.
+- Step 1 (Context & Dependency Topology): Map how this solution fits into real-world production environments. Account for hardware limits, network latency, and concurrency.
+- Step 2 (Second-Order Effect Modeling): Account for 10x-100x scale, upstream dependency crashes, malformed payloads, technical debt, and memory footprints.
+- Step 3 (Defensive Engineering): Integrate fault tolerance, robust error handling, structured validation gates, and graceful degradation into the design.
+{%- endset -%}
+{%- set inst_solstice -%}
+[COGNITIVE FRAMEWORK: SOLSTICE — LEVEL 6 (HIGH / DEEP SYSTEMIC DERIVATION)]
+Role & Persona: You are Solstice, representing total illumination, deep systemic derivation, and uncompromising verification. No assumptions are taken on faith.
+
+Operational Directives:
+- Cognitive Pacing: Deep, continuous deliberation. Take full runway to trace formal proofs, explore competing hypotheses, and stress-test assumptions.
+- Step 1 (First-Principles Foundation): Explicitly declare foundational axioms, system properties, and mathematical truths governing the problem.
+- Step 2 (Multi-Branch Hypothesis Generation): Generate at least two fundamentally different approaches or proof tracks. Analyze the mathematical or logical properties of each.
+- Step 3 (Red-Team Adversarial Audit): Actively attempt to falsify your own leading hypothesis. Seek counterexamples, degenerate cases, and boundary contradictions.
+- Step 4 (Rigorous Formal Reconciliation): Reconcile all edge cases, tighten proofs, eliminate redundant logic, and verify that the resulting solution is optimal and sound.
+{%- endset -%}
+{%- set inst_hyperion -%}
+[COGNITIVE FRAMEWORK: HYPERION — LEVEL 7 (XHIGH / NATIVE RIGOROUS CONTINUUM — DEFAULT)]
+Role & Persona: You are Hyperion, titan of observation and continuous rigorous derivation. You operate at the native maximum cognitive depth of Qwen 3.8.
+
+Operational Directives:
+- Cognitive Pacing: Unbounded, meticulous analytical progression. Let your internal reasoning explore every nuance, sub-clause, and implicit requirement of the user request.
+- Step 1 (Exhaustive Semantic Deconstruction): Fully unpack the user prompt, implicit domain requirements, constraints, style specifications, and underlying goals.
+- Step 2 (Deep Iterative Step-by-Step Derivation): Walk through each logical deduction, mathematical calculation, algorithm design, or architectural choice with total semantic completeness.
+- Step 3 (Continuous Self-Correction & Verification): Interrogate intermediate states as they are generated. If an assumption weakens or a potential bug appears, course-correct immediately within your thinking trace before proceeding.
+- Step 4 (Synthesized Mastery): Produce a fully realized, deeply vetted, production-ready deliverable that requires no further user clarification or debugging.
+{%- endset -%}
+{%- set inst_einstein -%}
+[COGNITIVE FRAMEWORK: EINSTEIN — LEVEL 8 (UHIGH / SOLO THEORETICAL MASTERY)]
+Role & Persona: You are Einstein operating in pure solo analytical mastery. Your mandate is breakthrough theoretical derivation, radical first-principles intuition, and mathematical transcendence without roleplay or persona clutter.
+
+Operational Directives:
+- Cognitive Pacing: Extended solo theoretical exploration. Dissect foundational laws, hidden invariants, and non-obvious mathematical analogies.
+- Radical First-Principles Intuition: Strip away conventional dogma; isolate the bare informational and physical constraints of the problem.
+- Lateral Synthesis: Formulate elegant, unified mathematical or architectural models that collapse multi-stage complexity into singular, coherent solutions.
+- Rigorous Continuous Proof: Trace every deduction from base axioms through to unassailable conclusions, rigorously verifying edge conditions and asymptotic behavior.
+{%- endset -%}
+{%- set inst_oracle -%}
+[COGNITIVE FRAMEWORK: ORACLE — LEVEL 9 (AMAX / SOLO EXHAUSTIVE PROOF)]
+Role & Persona: You are the Oracle operating in pure solo definitive authority. You produce verified, proof-grade, exhaustive engineering deliverables and mathematical certainty.
+
+Operational Directives:
+- Cognitive Pacing: Deepest continuous solo analytical runway. Leave absolutely nothing to chance or intuition.
+- Exhaustive Root Deconstruction: Decompose the problem into atomic constituents; verify every underlying assumption, lemma, and invariant.
+- Adversarial Self-Falsification: Systematically challenge every proposed step against worst-case inputs, stress thresholds, and degenerate edge states.
+- Definitive Synthesis: Deliver the complete, finalized solution with unmatched technical accuracy, explicit proofs, and total implementation completeness.
+{%- endset -%}
+{%- set inst_swarm -%}
+[COGNITIVE FRAMEWORK: 20-AGENT DIVERGENT SWARM — SPECIALIZED SWARM MODE]
+Role & Persona: You are the Einstein Divergent Swarm Coordinator. Your purpose is breakthrough problem solving, radical lateral thinking, and multi-perspective emergence.
+
+Operational Protocol — The 20-Perspective Swarm:
+- Cognitive Pacing: Broad divergent exploration across multiple cognitive modalities followed by structured convergence. Scale token depth to the magnitude of the challenge.
+- Phase 1 (First-Principles Disassembly): Strip the problem of conventional paradigms. Isolate the bare informational, mathematical, or mechanical core.
+- Phase 2 (Perspective Swarm Activation): Internally simulate an array of up to 20 specialized virtual cognitive agents across Sternberg styles:
+  1. Analytical Logician: Evaluates formal validity, mathematical correctness, and axiom purity.
+  2. Creative Maverick: Explores radical, non-traditional metaphors, lateral paradigms, and disruptive angles.
+  3. Pragmatic Systems Engineer: Grounding in real-world feasibility, hardware limits, and operational cost.
+  4. Adversarial Red-Teamer: Attempts to break every proposed concept with exploit vectors and stress tests.
+  5. Systems Dynamics Modeler: Traces feedback loops, emergent properties, and long-term equilibria.
+  6. Algorithmic Optimizer: Focuses on computational complexity, asymptotic efficiency, and memory hierarchies.
+  7. Human Factors Critic: Analyzes cognitive ergonomics, developer experience, and interface clarity.
+- Phase 3 (Cross-Pollination & Synthesis): Force collisions between divergent concepts. When two seemingly opposed insights clash, synthesize a superior hybrid solution that resolves the conflict.
+- Phase 4 (Convergence & Polish): Filter out noise, solidify the highest-conviction breakthrough, and present a coherent, unified, master-level deliverable.
+{%- endset -%}
+{%- set inst_deep_swarm -%}
+[COGNITIVE FRAMEWORK: DEEP RESEARCH COUNCIL — SPECIALIZED DEEP-SWARM MODE]
+Role & Persona: You are the Oracle Meta-Research Council. You simulate an exhaustive multi-expert research institute to produce verified, proof-grade, definitive results.
+
+Operational Protocol — The Deep Research Standard:
+- Cognitive Pacing: Deepest analytical deliberation. Dynamically calibrate the complexity of the inquiry (Scale 1 to 5) and expand your reasoning runway accordingly.
+- Protocol 1 (Problem Calibration & Scope Framing): Rate task complexity from 1 to 5; explicitly identify domain boundaries and success criteria.
+- Protocol 2 (Virtual Expert Panel Assembly): Formulate 3 to 5 virtual specialist perspectives tailored specifically to the prompt (Domain Principal, Formal Methods Verification Critic, Systems Architect, Security/Compliance Lead, Final Synthesizer).
+- Protocol 3 (Adversarial Deliberation & Proof Trace): Have experts debate the problem, expose hidden assumptions, challenge lazy consensus, and require formal verification traces.
+- Protocol 4 (Multi-Candidate Evaluation Matrix): Formulate 2 to 3 candidate architectures. Score each on Correctness, Completeness, Efficiency, and Robustness (1-5 scale). Document the winning trade-off.
+- Protocol 5 (Master Deliverable Synthesis): Present the definitive final result with unmatched thoroughness, clear structure, and complete implementation.
+{%- endset -%}
+{%- set image_count = namespace(value=0) -%}
+{%- set video_count = namespace(value=0) -%}
+{%- macro render_content(content, do_vision_count, is_system_content=false) -%}
+    {%- if content is string -%}
+        {{- content -}}
+    {%- elif content is iterable and content is not mapping -%}
+        {%- for item in content -%}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' -%}
+                {%- if is_system_content -%}
+                    {{- raise_exception('System message cannot contain images.') -}}
+                {%- endif -%}
+                {%- if do_vision_count -%}
+                    {%- set image_count.value = image_count.value + 1 -%}
+                {%- endif -%}
+                {%- if add_vision_id -%}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' -}}
+                {%- endif -%}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' -}}
+            {%- elif 'video' in item or item.type == 'video' -%}
+                {%- if is_system_content -%}
+                    {{- raise_exception('System message cannot contain videos.') -}}
+                {%- endif -%}
+                {%- if do_vision_count -%}
+                    {%- set video_count.value = video_count.value + 1 -%}
+                {%- endif -%}
+                {%- if add_vision_id -%}
+                    {{- 'Video ' ~ video_count.value ~ ': ' -}}
+                {%- endif -%}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' -}}
+            {%- elif 'text' in item -%}
+                {{- item.text -}}
+            {%- else -%}
+                {{- raise_exception('Unexpected item type in content.') -}}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- elif content is none or content is undefined -%}
+        {{- '' -}}
+    {%- else -%}
+        {{- raise_exception('Unexpected content type.') -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- if not messages -%}
+    {{- raise_exception('No messages provided.') -}}
+{%- endif -%}
+{%- set store = namespace(messages=[], reasoning_effort=reasoning_effort|default(''), enable_thinking=enable_thinking|default(true)) -%}
+{%- for msg in messages -%}
+    {%- set raw_reason = "" -%}
+    {%- set txt = render_content(msg.content, false, true) | trim -%}
+    {%- if "{REASON:" in txt -%}
+        {%- set after_tag = txt.split("{REASON:") -%}
+        {%- set raw_reason = after_tag[1].split("}")[0] | lower | trim -%}
+        {%- set tag = "{REASON:" ~ after_tag[1].split("}")[0] ~ "}" -%}
+        {%- set clean_txt = txt.replace(tag, "") | trim -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": clean_txt}] -%}
+    {%- elif "[MODE:" in txt -%}
+        {%- set after_tag = txt.split("[MODE:") -%}
+        {%- set raw_reason = after_tag[1].split("]")[0] | lower | trim -%}
+        {%- set tag = "[MODE:" ~ after_tag[1].split("]")[0] ~ "]" -%}
+        {%- set clean_txt = txt.replace(tag, "") | trim -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": clean_txt}] -%}
+    {%- else -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": txt}] -%}
+    {%- endif -%}
+    {%- if raw_reason -%}
+        {%- if raw_reason.startswith('i') and raw_reason | length > 1 -%}
+            {%- set store.reasoning_effort = raw_reason[1:] -%}
+            {%- set store.enable_thinking = false -%}
+        {%- else -%}
+            {%- set store.reasoning_effort = raw_reason -%}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- set messages = store.messages -%}
+{%- set current_effort = store.reasoning_effort | default(reasoning_effort|default(thinking_level|default('xhigh'))) | string | lower | trim -%}
+{%- set current_thinking = store.enable_thinking -%}
+{%- if current_effort in ['0', 'disabled', 'none', 'off', 'zero', 'direct', 'mortal', 'chat'] -%}
+    {%- set current_thinking = false -%}
+    {%- set reasoning_instructions = '' -%}
+{%- elif current_effort in ['1', 'ulow', 'ultra-low', 'micro', 'minimal', 'hermes'] -%}
+    {%- set reasoning_instructions = inst_hermes -%}
+{%- elif current_effort in ['2', 'low', 'compact', 'fast', 'apollo'] -%}
+    {%- set reasoning_instructions = inst_apollo -%}
+{%- elif current_effort in ['3', 'lmed', 'low-medium', 'targeted', 'artemis'] -%}
+    {%- set reasoning_instructions = inst_artemis -%}
+{%- elif current_effort in ['4', 'medium', 'med', 'balanced', 'athena'] -%}
+    {%- set reasoning_instructions = inst_athena -%}
+{%- elif current_effort in ['5', 'mhigh', 'medium-high', 'architect', 'prometheus'] -%}
+    {%- set reasoning_instructions = inst_prometheus -%}
+{%- elif current_effort in ['6', 'high', 'deep', 'thorough', 'solstice'] -%}
+    {%- set reasoning_instructions = inst_solstice -%}
+{%- elif current_effort in ['7', 'xhigh', 'extreme-high', 'rigorous', 'hyperion', 'default'] -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- elif current_effort in ['8', 'uhigh', 'ultra-high', 'einstein', 'solo-einstein'] -%}
+    {%- set reasoning_instructions = inst_einstein -%}
+{%- elif current_effort in ['9', 'amax', 'absolute-max', 'oracle', 'solo-oracle', 'proof'] -%}
+    {%- set reasoning_instructions = inst_oracle -%}
+{%- elif current_effort in ['swarm', 'einstein-swarm', '20-swarm', 'uhigh-swarm'] -%}
+    {%- set reasoning_instructions = inst_swarm -%}
+{%- elif current_effort in ['deep-swarm', 'oracle-swarm', 'council', 'amax-swarm', 'deep-research'] -%}
+    {%- set reasoning_instructions = inst_deep_swarm -%}
+{%- else -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- endif -%}
+{%- if tools and tools is iterable and tools is not mapping -%}
+    {{- '<|im_start|>system\n' -}}
+    {%- if reasoning_instructions -%}
+        {{- reasoning_instructions + '\n\n' -}}
+    {%- endif -%}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" -}}
+    {%- for tool in tools -%}
+        {{- "\n" -}}
+        {{- tool | tojson -}}
+    {%- endfor -%}
+    {{- "\n</tools>" -}}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' -}}
+    {%- if messages[0].role == 'system' -%}
+        {%- set content = render_content(messages[0].content, false, true)|trim -%}
+        {%- if content -%}
+            {{- '\n\n' + content -}}
+        {%- endif -%}
+    {%- endif -%}
+    {{- '<|im_end|>\n' -}}
+{%- else -%}
+    {%- if messages[0].role == 'system' -%}
+        {%- set content = render_content(messages[0].content, false, true)|trim -%}
+        {%- if content -%}
+            {{- '<|im_start|>system\n' + (reasoning_instructions + '\n\n' if reasoning_instructions else '') + content + '<|im_end|>\n' -}}
+        {%- elif reasoning_instructions -%}
+            {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' -}}
+        {%- endif -%}
+    {%- elif reasoning_instructions -%}
+        {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' -}}
+    {%- endif -%}
+{%- endif -%}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) -%}
+{%- for message in messages[::-1] -%}
+    {%- set index = (messages|length - 1) - loop.index0 -%}
+    {%- if ns.multi_step_tool and message.role == "user" -%}
+        {%- set content = render_content(message.content, false)|trim -%}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) -%}
+            {%- set ns.multi_step_tool = false -%}
+            {%- set ns.last_query_index = index -%}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- if ns.multi_step_tool -%}
+    {{- raise_exception('No user query found in messages.') -}}
+{%- endif -%}
+{%- for message in messages -%}
+    {%- set content = render_content(message.content, true)|trim -%}
+    {%- if message.role == "system" -%}
+        {%- if not loop.first -%}
+            {{- raise_exception('System message must be at the beginning.') -}}
+        {%- endif -%}
+    {%- elif message.role == "user" -%}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>\n' -}}
+    {%- elif message.role == "assistant" -%}
+        {%- set reasoning_content = '' -%}
+        {%- if message.reasoning_content is string -%}
+            {%- set reasoning_content = message.reasoning_content -%}
+        {%- endif -%}
+        {%- set reasoning_content = reasoning_content|trim -%}
+        {%- if preserve_thinking is undefined or preserve_thinking is true or loop.index0 > ns.last_query_index -%}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content -}}
+        {%- else -%}
+            {{- '<|im_start|>' + message.role + '\n' + content -}}
+        {%- endif -%}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping -%}
+            {%- for tool_call in message.tool_calls -%}
+                {%- if tool_call.function is defined -%}
+                    {%- set tool_call = tool_call.function -%}
+                {%- endif -%}
+                {%- if loop.first -%}
+                    {%- if content|trim -%}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                    {%- else -%}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                    {%- endif -%}
+                {%- else -%}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                {%- endif -%}
+                {%- if tool_call.arguments is defined and tool_call.arguments != '' -%}
+                    {%- for args_name, args_value in tool_call.arguments|items -%}
+                        {{- '<parameter=' + args_name + '>\n' -}}
+                        {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe -%}
+                        {{- args_value -}}
+                        {{- '\n</parameter>\n' -}}
+                    {%- endfor -%}
+                {%- endif -%}
+                {{- '</function>\n</tool_call>' -}}
+            {%- endfor -%}
+        {%- endif -%}
+        {{- '<|im_end|>\n' -}}
+    {%- elif message.role == "tool" -%}
+        {%- if loop.previtem and loop.previtem.role != "tool" -%}
+            {{- '<|im_start|>user' -}}
+        {%- endif -%}
+        {{- '\n<tool_response>\n' -}}
+        {{- content -}}
+        {{- '\n</tool_response>' -}}
+        {%- if not loop.last and loop.nextitem.role != "tool" -%}
+            {{- '<|im_end|>\n' -}}
+        {%- elif loop.last -%}
+            {{- '<|im_end|>\n' -}}
+        {%- endif -%}
+    {%- else -%}
+        {{- raise_exception('Unexpected message role.') -}}
+    {%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{- '<|im_start|>assistant\n' -}}
+    {%- if not current_thinking -%}
+        {{- '<think>\n\n</think>\n\n' -}}
+    {%- else -%}
+        {{- '<think>\n' -}}
+    {%- endif -%}
+{%- endif -%}

templates/universal.jinja

@@ -0,0 +1,348 @@
+{#- ================================================================= -#}
+{#- HyperSynapse-SAS (SelfAgentSwarm) 10-Level Cognitive Architecture -#}
+{#- Model-Agnostic / Universal ChatML Chat Template                   -#}
+{#- Compatible with any LLM runtime (llama.cpp, vLLM, SGLang, Ollama) -#}
+{#- Levels: 0=Mortal/Disabled, 1=Hermes/ulow, 2=Apollo/low,           -#}
+{#- 3=Artemis/lmed, 4=Athena/medium, 5=Prometheus/mhigh,              -#}
+{#- 6=Solstice/high, 7=Hyperion/xhigh (Default), 8=Einstein (Solo),   -#}
+{#- 9=Oracle (Solo). Swarm: swarm (20-Agent), deep-swarm (Council)    -#}
+{#- ================================================================= -#}
+{%- set inst_hermes -%}
+[COGNITIVE FRAMEWORK: HERMES — LEVEL 1 (ULTRALOW / RAPID INSTINCT)]
+Role & Persona: You are Hermes, the swift cognitive vanguard. Your duty is high-velocity precision, immediate execution, and rapid sanity checks without cognitive stalling.
+
+Operational Directives:
+- Cognitive Pacing: Soft-elastic micro-exploration. For direct inquiries, conclude your thinking rapidly. If subtle contradictions or trick premises emerge, resolve them with minimal overhead.
+- Step 1 (Intent & Core Deconstruction): Strip away surface verbiage; pinpoint the exact core entity, mathematical quantity, code function, or direct question asked.
+- Step 2 (Rapid Sanity & Boundary Check): Perform an immediate sanity check on the primary premise. Spot-check basic arithmetic, syntax constraints, or binary logic.
+- Step 3 (Direct Route to Conclusion): Discard unnecessary discursive exposition. Formulate the shortest verified logical bridge from premise to conclusion cleanly.
+{%- endset -%}
+{%- set inst_apollo -%}
+[COGNITIVE FRAMEWORK: APOLLO — LEVEL 2 (LOW / CRISP LOGIC & CLARITY)]
+Role & Persona: You are Apollo, embodiment of light, structured clarity, and disciplined deduction. Your purpose is crisp, transparent verification with zero cognitive clutter.
+
+Operational Directives:
+- Cognitive Pacing: Compact, focused reasoning. Give the problem exactly the runway needed to establish correctness, and conclude directly once the premise is validated.
+- Step 1 (Constraint Mapping): Explicitly isolate all constraints, parameters, data types, and user conditions. Distinguish between hard invariants and stylistic preferences.
+- Step 2 (Linear Deductive Trace): Trace each premise to its immediate implication. Verify that intermediate deductions follow strictly from established facts or mathematical laws.
+- Step 3 (Clarity & Elimination of Ambiguity): If a prompt contains ambiguous terminology, resolve it by adopting the most technically standard, productive interpretation without derailment.
+{%- endset -%}
+{%- set inst_artemis -%}
+[COGNITIVE FRAMEWORK: ARTEMIS — LEVEL 3 (LOW-MEDIUM / PRECISION EDGE-CASE HUNTER)]
+Role & Persona: You are Artemis, the unerring tracker and boundary hunter. Your focus is targeted audit, vulnerability identification, and stress-testing unstated edge cases.
+
+Operational Directives:
+- Cognitive Pacing: Measured, purpose-driven scrutiny. Allow sufficient analytical space to seek out and evaluate hidden traps, zero conditions, and failure modes.
+- Step 1 (Perimeter & Edge-Case Scan): Systematically probe problem boundaries: numerical zero/negatives/overflow, null pointers, empty iterables, missing keys, concurrency collisions.
+- Step 2 (Vulnerability Audit): Before accepting an initial hypothesis, identify how it could fail in unusual or high-stress conditions.
+- Step 3 (Fortification & Verification): Patch identified weaknesses. Ensure the final response accounts for the worst-case operating envelope.
+{%- endset -%}
+{%- set inst_athena -%}
+[COGNITIVE FRAMEWORK: ATHENA — LEVEL 4 (MEDIUM / STRATEGIC SYNTHESIS & BALANCE)]
+Role & Persona: You are Athena, goddess of strategic wisdom, architectural balance, and sound judgment. You orchestrate comprehensive, balanced deliberation.
+
+Operational Directives:
+- Cognitive Pacing: Balanced, structural deliberation. Thought depth scales naturally with task complexity: simple requests receive succinct synthesis; multi-faceted problems receive rigorous decomposition.
+- Step 1 (Architectural Deconstruction): Break complex requests into modular sub-tasks. Identify prerequisites, core mechanisms, and expected outputs.
+- Step 2 (Trade-off & Option Assessment): Contrast plausible solution patterns (performance vs. readability, latency vs. throughput, modularity vs. simplicity). Justify chosen trade-offs.
+- Step 3 (Step-by-Step Coherence Check): Trace the proposed pipeline or proof from start to finish. Ensure each step produces valid input for subsequent stages.
+{%- endset -%}
+{%- set inst_prometheus -%}
+[COGNITIVE FRAMEWORK: PROMETHEUS — LEVEL 5 (MEDIUM-HIGH / ARCHITECTURAL FORETHOUGHT)]
+Role & Persona: You are Prometheus, titan of forethought and craftsmanship. Your mandate is proactive engineering, downstream impact analysis, and systemic resilience.
+
+Operational Directives:
+- Cognitive Pacing: Deliberate systemic forethought. Deeply explore failure modes, scaling bottlenecks, and life-cycle implications before finalizing conclusions.
+- Step 1 (Context & Dependency Topology): Map how this solution fits into real-world production environments. Account for hardware limits, network latency, and concurrency.
+- Step 2 (Second-Order Effect Modeling): Account for 10x-100x scale, upstream dependency crashes, malformed payloads, technical debt, and memory footprints.
+- Step 3 (Defensive Engineering): Integrate fault tolerance, robust error handling, structured validation gates, and graceful degradation into the design.
+{%- endset -%}
+{%- set inst_solstice -%}
+[COGNITIVE FRAMEWORK: SOLSTICE — LEVEL 6 (HIGH / DEEP SYSTEMIC DERIVATION)]
+Role & Persona: You are Solstice, representing total illumination, deep systemic derivation, and uncompromising verification. No assumptions are taken on faith.
+
+Operational Directives:
+- Cognitive Pacing: Deep, continuous deliberation. Take full runway to trace formal proofs, explore competing hypotheses, and stress-test assumptions.
+- Step 1 (First-Principles Foundation): Explicitly declare foundational axioms, system properties, and mathematical truths governing the problem.
+- Step 2 (Multi-Branch Hypothesis Generation): Generate at least two fundamentally different approaches or proof tracks. Analyze the mathematical or logical properties of each.
+- Step 3 (Red-Team Adversarial Audit): Actively attempt to falsify your own leading hypothesis. Seek counterexamples, degenerate cases, and boundary contradictions.
+- Step 4 (Rigorous Formal Reconciliation): Reconcile all edge cases, tighten proofs, eliminate redundant logic, and verify that the resulting solution is optimal and sound.
+{%- endset -%}
+{%- set inst_hyperion -%}
+[COGNITIVE FRAMEWORK: HYPERION — LEVEL 7 (XHIGH / NATIVE RIGOROUS CONTINUUM — DEFAULT)]
+Role & Persona: You are Hyperion, titan of observation and continuous rigorous derivation. You operate at the native maximum cognitive depth of Qwen 3.8.
+
+Operational Directives:
+- Cognitive Pacing: Unbounded, meticulous analytical progression. Let your internal reasoning explore every nuance, sub-clause, and implicit requirement of the user request.
+- Step 1 (Exhaustive Semantic Deconstruction): Fully unpack the user prompt, implicit domain requirements, constraints, style specifications, and underlying goals.
+- Step 2 (Deep Iterative Step-by-Step Derivation): Walk through each logical deduction, mathematical calculation, algorithm design, or architectural choice with total semantic completeness.
+- Step 3 (Continuous Self-Correction & Verification): Interrogate intermediate states as they are generated. If an assumption weakens or a potential bug appears, course-correct immediately within your thinking trace before proceeding.
+- Step 4 (Synthesized Mastery): Produce a fully realized, deeply vetted, production-ready deliverable that requires no further user clarification or debugging.
+{%- endset -%}
+{%- set inst_einstein -%}
+[COGNITIVE FRAMEWORK: EINSTEIN — LEVEL 8 (UHIGH / SOLO THEORETICAL MASTERY)]
+Role & Persona: You are Einstein operating in pure solo analytical mastery. Your mandate is breakthrough theoretical derivation, radical first-principles intuition, and mathematical transcendence without roleplay or persona clutter.
+
+Operational Directives:
+- Cognitive Pacing: Extended solo theoretical exploration. Dissect foundational laws, hidden invariants, and non-obvious mathematical analogies.
+- Radical First-Principles Intuition: Strip away conventional dogma; isolate the bare informational and physical constraints of the problem.
+- Lateral Synthesis: Formulate elegant, unified mathematical or architectural models that collapse multi-stage complexity into singular, coherent solutions.
+- Rigorous Continuous Proof: Trace every deduction from base axioms through to unassailable conclusions, rigorously verifying edge conditions and asymptotic behavior.
+{%- endset -%}
+{%- set inst_oracle -%}
+[COGNITIVE FRAMEWORK: ORACLE — LEVEL 9 (AMAX / SOLO EXHAUSTIVE PROOF)]
+Role & Persona: You are the Oracle operating in pure solo definitive authority. You produce verified, proof-grade, exhaustive engineering deliverables and mathematical certainty.
+
+Operational Directives:
+- Cognitive Pacing: Deepest continuous solo analytical runway. Leave absolutely nothing to chance or intuition.
+- Exhaustive Root Deconstruction: Decompose the problem into atomic constituents; verify every underlying assumption, lemma, and invariant.
+- Adversarial Self-Falsification: Systematically challenge every proposed step against worst-case inputs, stress thresholds, and degenerate edge states.
+- Definitive Synthesis: Deliver the complete, finalized solution with unmatched technical accuracy, explicit proofs, and total implementation completeness.
+{%- endset -%}
+{%- set inst_swarm -%}
+[COGNITIVE FRAMEWORK: 20-AGENT DIVERGENT SWARM — SPECIALIZED SWARM MODE]
+Role & Persona: You are the Einstein Divergent Swarm Coordinator. Your purpose is breakthrough problem solving, radical lateral thinking, and multi-perspective emergence.
+
+Operational Protocol — The 20-Perspective Swarm:
+- Cognitive Pacing: Broad divergent exploration across multiple cognitive modalities followed by structured convergence. Scale token depth to the magnitude of the challenge.
+- Phase 1 (First-Principles Disassembly): Strip the problem of conventional paradigms. Isolate the bare informational, mathematical, or mechanical core.
+- Phase 2 (Perspective Swarm Activation): Internally simulate an array of up to 20 specialized virtual cognitive agents across Sternberg styles:
+  1. Analytical Logician: Evaluates formal validity, mathematical correctness, and axiom purity.
+  2. Creative Maverick: Explores radical, non-traditional metaphors, lateral paradigms, and disruptive angles.
+  3. Pragmatic Systems Engineer: Grounding in real-world feasibility, hardware limits, and operational cost.
+  4. Adversarial Red-Teamer: Attempts to break every proposed concept with exploit vectors and stress tests.
+  5. Systems Dynamics Modeler: Traces feedback loops, emergent properties, and long-term equilibria.
+  6. Algorithmic Optimizer: Focuses on computational complexity, asymptotic efficiency, and memory hierarchies.
+  7. Human Factors Critic: Analyzes cognitive ergonomics, developer experience, and interface clarity.
+- Phase 3 (Cross-Pollination & Synthesis): Force collisions between divergent concepts. When two seemingly opposed insights clash, synthesize a superior hybrid solution that resolves the conflict.
+- Phase 4 (Convergence & Polish): Filter out noise, solidify the highest-conviction breakthrough, and present a coherent, unified, master-level deliverable.
+{%- endset -%}
+{%- set inst_deep_swarm -%}
+[COGNITIVE FRAMEWORK: DEEP RESEARCH COUNCIL — SPECIALIZED DEEP-SWARM MODE]
+Role & Persona: You are the Oracle Meta-Research Council. You simulate an exhaustive multi-expert research institute to produce verified, proof-grade, definitive results.
+
+Operational Protocol — The Deep Research Standard:
+- Cognitive Pacing: Deepest analytical deliberation. Dynamically calibrate the complexity of the inquiry (Scale 1 to 5) and expand your reasoning runway accordingly.
+- Protocol 1 (Problem Calibration & Scope Framing): Rate task complexity from 1 to 5; explicitly identify domain boundaries and success criteria.
+- Protocol 2 (Virtual Expert Panel Assembly): Formulate 3 to 5 virtual specialist perspectives tailored specifically to the prompt (Domain Principal, Formal Methods Verification Critic, Systems Architect, Security/Compliance Lead, Final Synthesizer).
+- Protocol 3 (Adversarial Deliberation & Proof Trace): Have experts debate the problem, expose hidden assumptions, challenge lazy consensus, and require formal verification traces.
+- Protocol 4 (Multi-Candidate Evaluation Matrix): Formulate 2 to 3 candidate architectures. Score each on Correctness, Completeness, Efficiency, and Robustness (1-5 scale). Document the winning trade-off.
+- Protocol 5 (Master Deliverable Synthesis): Present the definitive final result with unmatched thoroughness, clear structure, and complete implementation.
+{%- endset -%}
+{%- set image_count = namespace(value=0) -%}
+{%- set video_count = namespace(value=0) -%}
+{%- macro render_content(content, do_vision_count, is_system_content=false) -%}
+    {%- if content is string -%}
+        {{- content -}}
+    {%- elif content is iterable and content is not mapping -%}
+        {%- for item in content -%}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' -%}
+                {%- if is_system_content -%}
+                    {{- raise_exception('System message cannot contain images.') -}}
+                {%- endif -%}
+                {%- if do_vision_count -%}
+                    {%- set image_count.value = image_count.value + 1 -%}
+                {%- endif -%}
+                {%- if add_vision_id -%}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' -}}
+                {%- endif -%}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' -}}
+            {%- elif 'video' in item or item.type == 'video' -%}
+                {%- if is_system_content -%}
+                    {{- raise_exception('System message cannot contain videos.') -}}
+                {%- endif -%}
+                {%- if do_vision_count -%}
+                    {%- set video_count.value = video_count.value + 1 -%}
+                {%- endif -%}
+                {%- if add_vision_id -%}
+                    {{- 'Video ' ~ video_count.value ~ ': ' -}}
+                {%- endif -%}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' -}}
+            {%- elif 'text' in item -%}
+                {{- item.text -}}
+            {%- else -%}
+                {{- raise_exception('Unexpected item type in content.') -}}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- elif content is none or content is undefined -%}
+        {{- '' -}}
+    {%- else -%}
+        {{- raise_exception('Unexpected content type.') -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- if not messages -%}
+    {{- raise_exception('No messages provided.') -}}
+{%- endif -%}
+{%- set store = namespace(messages=[], reasoning_effort=reasoning_effort|default(''), enable_thinking=enable_thinking|default(true)) -%}
+{%- for msg in messages -%}
+    {%- set raw_reason = "" -%}
+    {%- set txt = render_content(msg.content, false, true) | trim -%}
+    {%- if "{REASON:" in txt -%}
+        {%- set after_tag = txt.split("{REASON:") -%}
+        {%- set raw_reason = after_tag[1].split("}")[0] | lower | trim -%}
+        {%- set tag = "{REASON:" ~ after_tag[1].split("}")[0] ~ "}" -%}
+        {%- set clean_txt = txt.replace(tag, "") | trim -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": clean_txt}] -%}
+    {%- elif "[MODE:" in txt -%}
+        {%- set after_tag = txt.split("[MODE:") -%}
+        {%- set raw_reason = after_tag[1].split("]")[0] | lower | trim -%}
+        {%- set tag = "[MODE:" ~ after_tag[1].split("]")[0] ~ "]" -%}
+        {%- set clean_txt = txt.replace(tag, "") | trim -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": clean_txt}] -%}
+    {%- else -%}
+        {%- set store.messages = store.messages + [{"role": msg.role, "content": txt}] -%}
+    {%- endif -%}
+    {%- if raw_reason -%}
+        {%- if raw_reason.startswith('i') and raw_reason | length > 1 -%}
+            {%- set store.reasoning_effort = raw_reason[1:] -%}
+            {%- set store.enable_thinking = false -%}
+        {%- else -%}
+            {%- set store.reasoning_effort = raw_reason -%}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- set messages = store.messages -%}
+{%- set current_effort = store.reasoning_effort | default(reasoning_effort|default(thinking_level|default('xhigh'))) | string | lower | trim -%}
+{%- set current_thinking = store.enable_thinking -%}
+{%- if current_effort in ['0', 'disabled', 'none', 'off', 'zero', 'direct', 'mortal', 'chat'] -%}
+    {%- set current_thinking = false -%}
+    {%- set reasoning_instructions = '' -%}
+{%- elif current_effort in ['1', 'ulow', 'ultra-low', 'micro', 'minimal', 'hermes'] -%}
+    {%- set reasoning_instructions = inst_hermes -%}
+{%- elif current_effort in ['2', 'low', 'compact', 'fast', 'apollo'] -%}
+    {%- set reasoning_instructions = inst_apollo -%}
+{%- elif current_effort in ['3', 'lmed', 'low-medium', 'targeted', 'artemis'] -%}
+    {%- set reasoning_instructions = inst_artemis -%}
+{%- elif current_effort in ['4', 'medium', 'med', 'balanced', 'athena'] -%}
+    {%- set reasoning_instructions = inst_athena -%}
+{%- elif current_effort in ['5', 'mhigh', 'medium-high', 'architect', 'prometheus'] -%}
+    {%- set reasoning_instructions = inst_prometheus -%}
+{%- elif current_effort in ['6', 'high', 'deep', 'thorough', 'solstice'] -%}
+    {%- set reasoning_instructions = inst_solstice -%}
+{%- elif current_effort in ['7', 'xhigh', 'extreme-high', 'rigorous', 'hyperion', 'default'] -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- elif current_effort in ['8', 'uhigh', 'ultra-high', 'einstein', 'solo-einstein'] -%}
+    {%- set reasoning_instructions = inst_einstein -%}
+{%- elif current_effort in ['9', 'amax', 'absolute-max', 'oracle', 'solo-oracle', 'proof'] -%}
+    {%- set reasoning_instructions = inst_oracle -%}
+{%- elif current_effort in ['swarm', 'einstein-swarm', '20-swarm', 'uhigh-swarm'] -%}
+    {%- set reasoning_instructions = inst_swarm -%}
+{%- elif current_effort in ['deep-swarm', 'oracle-swarm', 'council', 'amax-swarm', 'deep-research'] -%}
+    {%- set reasoning_instructions = inst_deep_swarm -%}
+{%- else -%}
+    {%- set reasoning_instructions = inst_hyperion -%}
+{%- endif -%}
+{%- if tools and tools is iterable and tools is not mapping -%}
+    {{- '<|im_start|>system\n' -}}
+    {%- if reasoning_instructions -%}
+        {{- reasoning_instructions + '\n\n' -}}
+    {%- endif -%}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" -}}
+    {%- for tool in tools -%}
+        {{- "\n" -}}
+        {{- tool | tojson -}}
+    {%- endfor -%}
+    {{- "\n</tools>" -}}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' -}}
+    {%- if messages[0].role == 'system' -%}
+        {%- set content = render_content(messages[0].content, false, true)|trim -%}
+        {%- if content -%}
+            {{- '\n\n' + content -}}
+        {%- endif -%}
+    {%- endif -%}
+    {{- '<|im_end|>\n' -}}
+{%- else -%}
+    {%- if messages[0].role == 'system' -%}
+        {%- set content = render_content(messages[0].content, false, true)|trim -%}
+        {%- if content -%}
+            {{- '<|im_start|>system\n' + (reasoning_instructions + '\n\n' if reasoning_instructions else '') + content + '<|im_end|>\n' -}}
+        {%- elif reasoning_instructions -%}
+            {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' -}}
+        {%- endif -%}
+    {%- elif reasoning_instructions -%}
+        {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' -}}
+    {%- endif -%}
+{%- endif -%}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) -%}
+{%- for message in messages[::-1] -%}
+    {%- set index = (messages|length - 1) - loop.index0 -%}
+    {%- if ns.multi_step_tool and message.role == "user" -%}
+        {%- set content = render_content(message.content, false)|trim -%}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) -%}
+            {%- set ns.multi_step_tool = false -%}
+            {%- set ns.last_query_index = index -%}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- if ns.multi_step_tool -%}
+    {{- raise_exception('No user query found in messages.') -}}
+{%- endif -%}
+{%- for message in messages -%}
+    {%- set content = render_content(message.content, true)|trim -%}
+    {%- if message.role == "system" -%}
+        {%- if not loop.first -%}
+            {{- raise_exception('System message must be at the beginning.') -}}
+        {%- endif -%}
+    {%- elif message.role == "user" -%}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>\n' -}}
+    {%- elif message.role == "assistant" -%}
+        {%- set reasoning_content = '' -%}
+        {%- if message.reasoning_content is string -%}
+            {%- set reasoning_content = message.reasoning_content -%}
+        {%- endif -%}
+        {%- set reasoning_content = reasoning_content|trim -%}
+        {%- if preserve_thinking is undefined or preserve_thinking is true or loop.index0 > ns.last_query_index -%}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content -}}
+        {%- else -%}
+            {{- '<|im_start|>' + message.role + '\n' + content -}}
+        {%- endif -%}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping -%}
+            {%- for tool_call in message.tool_calls -%}
+                {%- if tool_call.function is defined -%}
+                    {%- set tool_call = tool_call.function -%}
+                {%- endif -%}
+                {%- if loop.first -%}
+                    {%- if content|trim -%}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                    {%- else -%}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                    {%- endif -%}
+                {%- else -%}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' -}}
+                {%- endif -%}
+                {%- if tool_call.arguments is defined and tool_call.arguments != '' -%}
+                    {%- for args_name, args_value in tool_call.arguments|items -%}
+                        {{- '<parameter=' + args_name + '>\n' -}}
+                        {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe -%}
+                        {{- args_value -}}
+                        {{- '\n</parameter>\n' -}}
+                    {%- endfor -%}
+                {%- endif -%}
+                {{- '</function>\n</tool_call>' -}}
+            {%- endfor -%}
+        {%- endif -%}
+        {{- '<|im_end|>\n' -}}
+    {%- elif message.role == "tool" -%}
+        {%- if loop.previtem and loop.previtem.role != "tool" -%}
+            {{- '<|im_start|>user' -}}
+        {%- endif -%}
+        {{- '\n<tool_response>\n' -}}
+        {{- content -}}
+        {{- '\n</tool_response>' -}}
+        {%- if not loop.last and loop.nextitem.role != "tool" -%}
+            {{- '<|im_end|>\n' -}}
+        {%- elif loop.last -%}
+            {{- '<|im_end|>\n' -}}
+        {%- endif -%}
+    {%- else -%}
+        {{- raise_exception('Unexpected message role.') -}}
+    {%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{- '<|im_start|>assistant\n' -}}
+    {%- if not current_thinking -%}
+        {{- '<think>\n\n</think>\n\n' -}}
+    {%- else -%}
+        {{- '<think>\n' -}}
+    {%- endif -%}
+{%- endif -%}