Skip to content

CLAUDE.md

Latest commit

 

History

History
102 lines (71 loc) · 4.66 KB

CLAUDE.md

File metadata and controls

102 lines (71 loc) · 4.66 KB
Raw

CLAUDE.md

Testing

Run python -m pytest tests/ -v before committing. All tests must pass. Tests run inside Docker: docker compose exec app python -m pytest tests/ -v (may need pip install pytest pytest-asyncio first if the container was rebuilt).

Running the Agent Simulation

The simulation runs in a one-off container named agent-run:

# Resume an existing run (no budget limit):
docker compose --profile agent run -d --name agent-run agent python -m src.agent.main --budget 0

# Resume with a budget cap (e.g. 50 LLM calls per agent):
docker compose --profile agent run -d --name agent-run agent python -m src.agent.main --budget 50

# Fresh run (wipes agent_messages/channels, keeps proposals):
docker compose --profile agent run -d --name agent-run agent python -m src.agent.main --fresh --budget 0

# With a time limit (minutes):
docker compose --profile agent run -d --name agent-run agent python -m src.agent.main --max-runtime 60 --budget 0

Before restarting, always save logs and rebuild containers:

# 1. Save logs
docker logs agent-run > logs/run_$(date +%s).log 2>&1
ls -t logs/run_*.log | tail -n +11 | xargs rm -f

# 2. Stop the old container — GRACEFULLY. `docker rm -f` sends SIGKILL, which
#    skips the shutdown flush and permanently loses the in-flight turn's
#    messages (the DB, not Slack, is the durable store). `docker stop` sends
#    SIGTERM; -t 30 leaves room for an in-flight LLM call to finish.
docker stop -t 30 agent-run
docker rm agent-run

# 3. Rebuild app + worker (picks up code changes)
docker compose up -d --build app worker

# 4. Start the new run
docker compose --profile agent run -d --name agent-run agent python -m src.agent.main --budget 0

Note: The agent-run container uses mounted source code but the Python process only loads modules at startup. Code changes require a container restart to take effect. After any code change that affects the running agent process, flag this to the user so they can decide whether to restart. (Roster changes — activating/inactivating agents or setting a new slack_bot_token in AgentRegistry — do NOT need a restart; they're picked up live by _sync_roster_from_db.)

Adding New PIs

The AgentRegistry table is the single source of truth for the agent roster. There is no longer a PILOT_LABS list and no per-agent config.py token fields to edit. A running agent-run re-syncs the roster from the DB every ~30s (_sync_roster_from_db), so flipping an agent to status='active' (with a token on its row) makes it go live without a restart.

1. Create user records and generate profiles

Look up each PI's ORCID ID (search orcid.org or the ORCID public API). Add them to orcids.txt with a comment line, then seed:

docker compose exec app python -m src.cli seed-profiles --file new_orcids.txt

This creates User rows and enqueues profile generation jobs (processed by the worker).

2. Create agent registry entries

Each agent needs an AgentRegistry row with a unique agent_id (lowercase last name) and bot_name ({LastName}Bot), created status='pending'. Self-service signups (src/routers/agent_page.py) and the backfill scripts both create these automatically.

Last-name collisions: If a last name is already taken (e.g., Chunlei Wu = wu), prefix with the first initial (e.g., Peng Wu = pwu / PWuBot). The web UI applies this logic automatically.

3. Provision the Slack bot + activate (admin UI)

Go to /admin/agents → the pending agent → Provision. This creates the Slack app via the Manifest API and sends you to Slack's install screen; on approval you return to the page with the bot token filled in and saved to AgentRegistry.slack_bot_token. Click Approve & Activate to set status='active'. The running simulation picks the agent up on its next roster sync — no .env edit, no config.py edit, no restart.

Requires SLACK_CONFIG_TOKEN / SLACK_CONFIG_REFRESH_TOKEN in the environment (the rotating pair is persisted in the app_settings KV table) and a public base_url.

Bulk provisioning (many agents at once) still uses the host script. First export the roster from the container, then run the script on the host:

docker exec copi-python-app-1 python scripts/export_agent_roster.py   # writes data/agent_roster.json
python3 scripts/provision_slack_bots.py                               # host: creates apps, prints OAuth URLs

The host script writes tokens to .env; import them into the DB column with:

docker exec copi-python-app-1 python scripts/backfill_agent_tokens.py

(.env + config.py get_slack_tokens() remain a read fallback, but the DB column is authoritative.)