egregore
ebd78cf88e
- Uses slowapi for rate limiting - Default: 10 requests/minute per IP - Configurable via RATE_LIMIT env var - Returns HTTP 429 when limit exceeded
102 lines
2.6 KiB
Python
102 lines
2.6 KiB
Python
#!/usr/bin/env python3
|
|
"""
|
|
Egregore Reason Service - AI reasoning API
|
|
|
|
Provides HTTP API for conversation processing with Claude.
|
|
Runs on port 8081.
|
|
"""
|
|
|
|
import os
|
|
from typing import Optional
|
|
|
|
from fastapi import FastAPI, HTTPException, Request
|
|
from pydantic import BaseModel
|
|
from dotenv import load_dotenv
|
|
from slowapi import Limiter, _rate_limit_exceeded_handler
|
|
from slowapi.util import get_remote_address
|
|
from slowapi.errors import RateLimitExceeded
|
|
import anthropic
|
|
|
|
from tools import TOOLS, execute_tool
|
|
from prompts import get_system_prompt
|
|
from conversation import process_conversation
|
|
|
|
# Load environment
|
|
load_dotenv("/home/admin/.env")
|
|
|
|
ANTHROPIC_API_KEY = os.getenv("ANTHROPIC_API_KEY")
|
|
if not ANTHROPIC_API_KEY:
|
|
raise ValueError("ANTHROPIC_API_KEY not set")
|
|
|
|
# Rate limiting configuration
|
|
RATE_LIMIT = os.getenv("RATE_LIMIT", "10/minute")
|
|
|
|
# Initialize rate limiter
|
|
limiter = Limiter(key_func=get_remote_address)
|
|
|
|
app = FastAPI(title="Egregore Reason Service", docs_url="/docs")
|
|
app.state.limiter = limiter
|
|
app.add_exception_handler(RateLimitExceeded, _rate_limit_exceeded_handler)
|
|
|
|
# Initialize Anthropic client
|
|
client = anthropic.AsyncAnthropic(api_key=ANTHROPIC_API_KEY)
|
|
|
|
|
|
# Request models
|
|
class ProcessRequest(BaseModel):
|
|
model: str = "claude-sonnet-4-20250514"
|
|
history: list # Conversation history in Claude API format
|
|
max_iterations: int = 10
|
|
|
|
|
|
class ToolRequest(BaseModel):
|
|
name: str
|
|
input: dict
|
|
|
|
|
|
# Endpoints
|
|
@app.post("/process")
|
|
@limiter.limit(RATE_LIMIT)
|
|
async def api_process(request: Request, req: ProcessRequest):
|
|
"""Process a conversation with tool use loop"""
|
|
try:
|
|
response_blocks = await process_conversation(
|
|
client=client,
|
|
model=req.model,
|
|
history=req.history,
|
|
max_iterations=req.max_iterations
|
|
)
|
|
return {"blocks": response_blocks}
|
|
except anthropic.APIError as e:
|
|
raise HTTPException(status_code=500, detail=str(e))
|
|
|
|
|
|
@app.post("/tool")
|
|
async def api_execute_tool(req: ToolRequest):
|
|
"""Execute a single tool directly"""
|
|
result = await execute_tool(req.name, req.input)
|
|
return {"result": result}
|
|
|
|
|
|
@app.get("/tools")
|
|
async def api_get_tools():
|
|
"""Get available tool definitions"""
|
|
return {"tools": TOOLS}
|
|
|
|
|
|
@app.get("/prompt")
|
|
async def api_get_prompt():
|
|
"""Get current system prompt with context"""
|
|
prompt = await get_system_prompt()
|
|
return {"prompt": prompt}
|
|
|
|
|
|
@app.get("/health")
|
|
async def health():
|
|
"""Health check endpoint"""
|
|
return {"status": "ok", "service": "reason"}
|
|
|
|
|
|
if __name__ == "__main__":
|
|
import uvicorn
|
|
uvicorn.run(app, host="127.0.0.1", port=8081)
|