Files
Cortex-Inara/cortex/routers/orchestrator.py
Scott Idem ed472ce9a0 feat: Intelligence Layer Phase 1 — orchestrator service
Adds the Gemini API orchestrator (ReAct tool loop → Claude responder):

Orchestrator engine + router:
- orchestrator_engine.py: Gemini API tool loop, Claude CLI handoff
- routers/orchestrator.py: POST /orchestrate (async job queue), GET /orchestrate/{job_id}

Tools (cortex/tools/):
- web.py: DuckDuckGo web search (no key required)
- ae_knowledge.py: ae_journal_search + ae_journal_entry_create (AE V3 API)
- ae_tasks.py: ae_task_list (reads agents_sync Kanban filesystem)
- files.py: file_read (path-allowlisted to safe dirs)

Config + deps:
- config.py: orchestrator, DuckDuckGo, and AE API settings
- requirements.txt: google-genai, duckduckgo-search
- .env.default: reference config with all new keys documented

Docs:
- CLAUDE.md, README.md, documentation/ added to repo
- Port references updated 7331 → 8000 throughout
- Default model updated to gemini-2.5-flash

Tested: ae_task_list, ae_journal_search, web_search all working end-to-end.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-03-18 19:37:49 -04:00

175 lines
5.8 KiB
Python

"""
Orchestrator router — POST /orchestrate, GET /orchestrate/{job_id}
Accepts a task description, runs it through the orchestrator engine
(Gemini tool loop → Claude response), and returns the result.
Designed to be triggered from:
- The Cortex web UI (future "Agent mode" toggle)
- Cron jobs: curl -X POST http://localhost:8000/orchestrate -d '{"task":"..."}'
- Webhooks: Gitea, Aether events, etc.
"""
import asyncio
import logging
import uuid
from datetime import datetime, timezone
from fastapi import APIRouter
from pydantic import BaseModel
from config import settings
from context_loader import load_context
import orchestrator_engine
logger = logging.getLogger(__name__)
router = APIRouter(prefix="/orchestrate", tags=["orchestrator"])
# ---------------------------------------------------------------------------
# In-memory job store
# Jobs are keyed by UUID. For this phase, memory is fine — jobs are short-lived.
# ---------------------------------------------------------------------------
_jobs: dict[str, dict] = {}
_jobs_lock = asyncio.Lock()
# ---------------------------------------------------------------------------
# Request / response models
# ---------------------------------------------------------------------------
class OrchestrateRequest(BaseModel):
task: str
session_id: str | None = None # include session history in context
tier: int | None = None # Inara context tier (default from settings)
respond_with_claude: bool = True # False = return Gemini summary only (faster, for cron)
include_long: bool = True
include_mid: bool = True
include_short: bool = True
class OrchestrateResponse(BaseModel):
job_id: str
status: str # "queued" | "running" | "complete" | "error"
class JobStatusResponse(BaseModel):
job_id: str
status: str
task: str
created_at: str
completed_at: str | None = None
response: str | None = None
tool_calls: list[dict] | None = None
backend: str | None = None
gemini_summary: str | None = None
error: str | None = None
# ---------------------------------------------------------------------------
# Endpoints
# ---------------------------------------------------------------------------
@router.post("", response_model=OrchestrateResponse)
async def orchestrate(req: OrchestrateRequest) -> OrchestrateResponse:
"""Submit a task to the orchestrator. Returns a job_id to poll."""
job_id = str(uuid.uuid4())
now = datetime.now(timezone.utc).isoformat()
job: dict = {
"job_id": job_id,
"status": "queued",
"task": req.task,
"created_at": now,
"completed_at": None,
"response": None,
"tool_calls": None,
"backend": None,
"gemini_summary": None,
"error": None,
}
async with _jobs_lock:
_jobs[job_id] = job
# Run in background — caller polls GET /orchestrate/{job_id}
asyncio.create_task(_run_job(job_id, req))
logger.info("Orchestrator job queued: %s%.80s", job_id, req.task)
return OrchestrateResponse(job_id=job_id, status="queued")
@router.get("/{job_id}", response_model=JobStatusResponse)
async def job_status(job_id: str) -> JobStatusResponse:
"""Poll the status of an orchestrator job."""
async with _jobs_lock:
job = _jobs.get(job_id)
if job is None:
from fastapi import HTTPException
raise HTTPException(status_code=404, detail=f"Job {job_id} not found")
return JobStatusResponse(**job)
@router.get("", response_model=list[JobStatusResponse])
async def list_jobs() -> list[JobStatusResponse]:
"""List all jobs (most recent first). Useful for debugging."""
async with _jobs_lock:
jobs = sorted(_jobs.values(), key=lambda j: j["created_at"], reverse=True)
return [JobStatusResponse(**j) for j in jobs]
# ---------------------------------------------------------------------------
# Background runner
# ---------------------------------------------------------------------------
async def _run_job(job_id: str, req: OrchestrateRequest) -> None:
"""Execute the orchestration job and update the job store."""
async with _jobs_lock:
_jobs[job_id]["status"] = "running"
try:
# Load Inara's system prompt (same as the chat router does)
tier = req.tier or settings.default_tier
system_prompt = load_context(
tier,
include_long=req.include_long,
include_mid=req.include_mid,
include_short=req.include_short,
)
# Load session history if a session_id was provided
session_messages: list[dict] | None = None
if req.session_id:
from session_store import load as load_session
session_messages = load_session(req.session_id) or None
result = await orchestrator_engine.run(
task=req.task,
system_prompt=system_prompt,
session_messages=session_messages,
respond_with_claude=req.respond_with_claude,
)
now = datetime.now(timezone.utc).isoformat()
async with _jobs_lock:
_jobs[job_id].update({
"status": "complete",
"completed_at": now,
"response": result.response,
"tool_calls": result.tool_calls,
"backend": result.backend,
"gemini_summary": result.gemini_summary,
})
logger.info("Orchestrator job complete: %s (%d tool calls)", job_id, len(result.tool_calls))
except Exception as e:
logger.exception("Orchestrator job failed: %s", job_id)
now = datetime.now(timezone.utc).isoformat()
async with _jobs_lock:
_jobs[job_id].update({
"status": "error",
"completed_at": now,
"error": str(e),
})