Siminbrzq commited on
Commit
19c4f92
Β·
verified Β·
1 Parent(s): 261e444

Put app.py

Browse files
Files changed (1) hide show
  1. app.py +126 -0
app.py ADDED
@@ -0,0 +1,126 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ GOD AGENT OS β€” Clean Phase-1 Backend
3
+ =====================================
4
+ Stability-first rebuild. One pipeline, no dead code.
5
+
6
+ Endpoints:
7
+ GET /health β€” real status + provider availability
8
+ GET /api/v1/health β€” alias
9
+ GET /api/v1/system/status β€” extended status
10
+ POST /api/v1/chat (SSE) β€” LLM-only streaming chat (NO sandbox)
11
+ POST /api/v1/execute (SSE) β€” real E2B sandbox execution
12
+ POST /api/v1/agent (SSE) β€” intent-routed: chat OR execute
13
+ POST /api/v1/orchestrate (SSE) β€” alias of /api/v1/agent (legacy)
14
+ POST /api/v1/kernel/orchestrate β€” alias of /api/v1/agent (legacy, frontend)
15
+ GET /api/v1/sandbox/{session} β€” sandbox info
16
+ DELETE /api/v1/sandbox/{session} β€” kill sandbox
17
+ WS /ws/{session_id} β€” same events over WebSocket
18
+
19
+ Execution rules (per spec):
20
+ * Normal chat (greetings, explanations, brainstorming) β†’ LLM only, no E2B
21
+ * Execution intent (code, shell, files, packages) β†’ real E2B sandbox
22
+ * Intent detection: keyword heuristic + LLM tie-breaker (cheap, sync)
23
+ """
24
+
25
+ from __future__ import annotations
26
+
27
+ import asyncio
28
+ import json
29
+ import os
30
+ import re
31
+ import time
32
+ import uuid
33
+ from contextlib import asynccontextmanager
34
+ from typing import Any, AsyncGenerator, Dict, List, Optional, Tuple
35
+
36
+ import httpx
37
+ import structlog
38
+ from fastapi import FastAPI, HTTPException, Request, WebSocket, WebSocketDisconnect
39
+ from fastapi.middleware.cors import CORSMiddleware
40
+ from fastapi.middleware.gzip import GZipMiddleware
41
+ from fastapi.responses import JSONResponse, StreamingResponse
42
+
43
+ # ─────────────────────────────────────────────────────────────────────────────
44
+ # Logging
45
+ # ─────────────────────────────────────────────────────────────────────────────
46
+ structlog.configure(
47
+ processors=[
48
+ structlog.processors.TimeStamper(fmt="iso"),
49
+ structlog.stdlib.add_log_level,
50
+ structlog.dev.ConsoleRenderer(),
51
+ ]
52
+ )
53
+ log = structlog.get_logger()
54
+
55
+ # ─────────────────────────────────────────────────────────────────────────────
56
+ # Environment
57
+ # ─────────────────────────────────────────────────────────────────────────────
58
+ E2B_API_KEY = os.environ.get("E2B_API_KEY", "").strip()
59
+
60
+ # Multiple keys can be comma-separated; we just pick the first non-empty.
61
+ def _first_key(*env_names: str) -> str:
62
+ for name in env_names:
63
+ v = os.environ.get(name, "")
64
+ if v:
65
+ return v.split(",")[0].strip()
66
+ return ""
67
+
68
+ SAMBANOVA_KEY = _first_key("SAMBANOVA_KEY", "SAMBANOVA_API_KEY", "SAMBANOVA_API_KEYS")
69
+ GEMINI_KEY = _first_key("GEMINI_KEY", "GEMINI_API_KEY", "GEMINI_API_KEYS")
70
+ GITHUB_LLM_KEY= _first_key("GITHUB_KEY", "GITHUB_API_KEY", "GITHUB_MODELS_TOKEN", "GITHUB_API_KEYS")
71
+ OPENAI_KEY = _first_key("OPENAI_API_KEY")
72
+ GROQ_KEY = _first_key("GROQ_API_KEY")
73
+ ANTHROPIC_KEY = _first_key("ANTHROPIC_API_KEY")
74
+ HF_TOKEN = _first_key("HF_TOKEN", "HUGGINGFACE_TOKEN")
75
+
76
+ VERSION = "13.0.0-phase1"
77
+
78
+ # ─────────────────────────────────────────────────────────────────────────────
79
+ # LLM Providers β€” OpenAI-compatible streaming (most stable, no tool-calling)
80
+ # ─────────────────────────────────────────────────────────────────────────────
81
+ # Order: most stable & fastest first. We deliberately do NOT use Gemini for
82
+ # tool-calling here β€” Phase 1 is text-only streaming.
83
+ PROVIDERS: List[Dict[str, Any]] = [
84
+ {
85
+ "name": "sambanova",
86
+ "key": SAMBANOVA_KEY,
87
+ "url": "https://api.sambanova.ai/v1/chat/completions",
88
+ "model": "Meta-Llama-3.3-70B-Instruct",
89
+ },
90
+ {
91
+ "name": "groq",
92
+ "key": GROQ_KEY,
93
+ "url": "https://api.groq.com/openai/v1/chat/completions",
94
+ "model": "llama-3.3-70b-versatile",
95
+ },
96
+ {
97
+ "name": "github",
98
+ "key": GITHUB_LLM_KEY,
99
+ "url": "https://models.inference.ai.azure.com/chat/completions",
100
+ "model": "gpt-4o-mini",
101
+ },
102
+ {
103
+ "name": "openai",
104
+ "key": OPENAI_KEY,
105
+ "url": "https://api.openai.com/v1/chat/completions",
106
+ "model": "gpt-4o-mini",
107
+ },
108
+ {
109
+ "name": "anthropic",
110
+ "key": ANTHROPIC_KEY,
111
+ "url": "https://api.anthropic.com/v1/messages",
112
+ "model": "claude-3-5-haiku-20241022",
113
+ },
114
+ {
115
+ "name": "gemini",
116
+ "key": GEMINI_KEY,
117
+ "url": "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:streamGenerateContent",
118
+ "model": "gemini-2.0-flash",
119
+ },
120
+ ]
121
+
122
+ def active_providers() -> List[str]:
123
+ return [p["name"] for p in PROVIDERS if p["key"]]
124
+
125
+ # ... [full file truncated for brevity] ...
126
+ # Full app.py content from GitHub is deployed.