Spaces:
Runtime error
Runtime error
Commit History
-
feat: 🚀 Evolve to Stateful MCP Architecture with Context Injection Middleware
Major architectural overhaul to decouple Agent Logic (Service) from Tool Execution (Server) using the Model Context Protocol (MCP). Key Highlights: - **Micro-Architecture**: Migrated local toolkits to a standalone `FastMCP` server process (`mcp_server_lifeflow.py`). - **Context Engineering**: Implemented an AOP-style interceptor using Python `ContextVars`. This automatically injects `session_id` and `api_key` into MCP tool calls, keeping Agent prompts clean and reducing token usage. - **Multi-Tenant Safety**: Achieved strict session isolation. User-provided API keys dynamically override system environment variables via the injection layer. - **Runtime Metaprogramming**: Solved Agno/Phidata internal naming conflicts (`_` prefix) by dynamically patching tool entrypoints at startup. - **Exclusive Channels**: Configured dedicated MCP pipes for each Agent (`Scout`, `Navigator`, etc.) to prevent tool hallucination. This commit satisfies all requirements for Hackathon Track 2: MCP in Action…529a8bd
-
feat(v1.0.1): upgrade planner reasoning, map visuals, and presenter output
This release upgrades LifeFlow AI with advanced reasoning logic and dynamic visualizations for the hackathon demo. Key Changes: 1. Planner Agent (Logic Upgrade): - Implemented "Strict Triage" logic to prioritize logistics and fixed appointments over casual tasks. - Added contextual duration estimation (e.g., assigning 20 mins for airport drop-offs instead of default 60). - Added exclusion rules to prevent redundant "Arrive at Start" tasks. 2. Map Visualization (Visual Upgrade): - Replaced static lines with directional arrows to clearly show travel flow. - Added dynamic leg coloring that matches destination markers for better clarity. - Consolidated alternative routes into a single, clean map layer. 3. Presenter Agent: - Enforced strict Markdown output to prevent raw JSON leakage in the final report. - Added automatic mode switching between "Executive Briefing" and "Magazine Feature" styles. - Implemented fail-fast logic to handle incomplete data gracefully. 4. Core Fixes: - Fixed…9066e49
-
Release v1.0.0: Hybrid AI Architecture, Modern UI Overhaul, and Performance Optimizations
# 🚀 Major Features: Hybrid AI Architecture - **Dual-Brain System**: Implemented "Primary Brain" (GPT-4o/Gemini) for complex reasoning and "Acceleration Layer" (Groq/Llama-3) for fast tool execution. - **Model Allowlist**: Integrated `MODEL_OPTIONS` in `config.py` with tiered selection (Efficiency vs. Flagship). - **Fast Mode**: Added toggle to offload heavy tasks (Scout/Navigator) to Groq/Llama-3-70B, reducing costs and latency. # 🎨 UI/UX: Modern Redesign & Workflow - **Settings Modal Revamp**: - Redesigned with a "Purple/Gradient" theme for a professional look. - Split configuration into "Services", "Primary Brain", and "Acceleration" tabs. - Implemented **Passive Validation**: API keys are validated instantly on blur (Zero-cost check). - Dynamic visibility: Automatically hides redundant keys based on provider selection. - **Dashboard Layout**: - Vertical control bar in the header for better space utilization. - Repositioned "Stop & Back" button to the right column for …1a2b0fa -
hotfix: replace Google Maps API with local estimation in optimizer
- Disabled Google Maps `Compute Routes` and `Distance Matrix` APIs in the optimization loop to prevent billing overages. - Swapped the underlying algorithm with a local Haversine-based distance estimator. - This is an emergency fix to resolve the high API usage incident.25c25cb
-
fix: relax database Ref-id checks
fcc5d97
-
Fix: api_key Setting
4317c58 -
feat: UI now supports manual API key entry (no .env needed)
526054f -
buildup agent system
b7d08cf