aipm

An automatic, suggest-only work bot. It watches your work threads — GitHub issues and PRs, Slack threads — figures out when someone owes an action or a thread has gone stale, and drafts a targeted nudge to the right person's Slack DM. It also keeps a per-thread working-notes summary up to date. It never acts on its own: it drafts, a human approves.

The problem it solves: the useful chores (nudging a stalled review, noticing a PR with no reviewer, summarizing where a thread actually stands) are easy to forget, and platform notifications are too noisy to be the signal. This bot makes those chores happen on their own, and pushes a small number of high-value nudges instead of a firehose.

Principles

• Suggest-only. The bot owns and edits its own artifacts (its working-notes comment, its DMs). Anything touching human-authored content is a proposal approved with one reaction.
• Low-noise. One nudge per (person, thread, signal) per quiet period, then it falls back to a digest. Mute/snooze always wins.
• Deterministic where it matters. Detecting that an action is owed is plain logic over the thread timeline. The LLM is used only for judgment and wording (did this reply actually answer the question? how should the nudge read?).
• Shadow mode first. Compute everything, post nothing, log what it would do. Review the log, then enable posting one capability at a time.

Platforms

GitHub and Slack are adapters behind a common interface; the LLM provider is an adapter too. Adding a platform means implementing the adapter, not touching the engine.

Stack

Cloudflare-native: Workers (webhook + Slack ingress) · Cron Triggers (staleness sweeps) · Queues (bound + retry LLM work) · Durable Objects (serialize per-thread updates, never double-nudge) · D1 (relational state) · KV (delivery-id dedupe) · Workers AI behind AI Gateway (caching + swappable provider). Deploy with Wrangler.

Quickstart

# 1. Install and test
pnpm install
pnpm --filter @aipm/worker test

# 2. Create the GitHub App (issues/PRs RW, contents R, members R; org-wide install)
#    Subscribe to: issues, issue_comment, pull_request, pull_request_review,
#    pull_request_review_comment, pull_request_review_thread

# 3. Point both webhooks at the Worker, then run locally
pnpm --filter @aipm/worker exec wrangler dev

Deployment config

apps/worker/wrangler.jsonc is source-level scaffolding for local development, tests, and type generation. It intentionally does not contain a real Cloudflare account id, D1 database id, or KV namespace ids.

Production deployment is owned outside this source repo. A deployment repo or CI job should:

1. Check out gantrydev/aipm at the desired ref.
2. Provide a real apps/worker/wrangler.jsonc for that environment.
3. Apply D1 migrations from apps/worker/migrations.
4. Run wrangler deploy.
5. Provide runtime secrets with Wrangler or the deployer's secret-management workflow.

This keeps account ids, resource ids, rosters, and secret policy in the private infra layer that owns the deployment.

To customize the LLM summary prompts (NOTES_PROMPT / CLUSTER_PROMPT), see docs/configuring-prompts.md.

See DESIGN.md for the architecture.