A 4-agent crew for solo OSS maintainers โ triages issues, reviews PRs, writes changelogs, keeps docs current. Built for the solo-maintainer-burnout problem.
Solo OSS maintainers burn out on triage, not code. A repo with 200 stars and one maintainer accumulates issues faster than the maintainer can read them, and PRs sit unreviewed for weeks because the cognitive load of "is this PR safe?" is worse than writing the code yourself. Most popular OSS projects either grow a maintainer team or quietly stagnate.
This 4-agent GitHub maintainer team is built for the solo-maintainer-burnout problem. Triager labels and dedupes incoming issues. Critic does first-pass PR review and leaves comments before you look, so when you do review you start from "is Critic right?" instead of "what does this PR do?". Logger writes changelogs and proposes semver bumps. Docs keeps the README in sync with what the code actually does. The crew won't replace your judgment on hard architectural calls, but it will absorb the 80% of repo work that's pattern-matching.
Labels, reproduces, and prioritizes incoming issues. Closes duplicates with a link to the existing thread.
First-pass PR review โ tests, style, scope creep, breaking changes. Leaves comments before you look.
Generates changelog entries from merged PRs; semver-bumps the version.
Updates README, API docs, and example snippets when public surface changes.
Triager watches the issues feed: labels new issues by type/priority, asks for repro steps if missing, closes obvious dupes.
Critic watches PRs: runs tests, comments on style and scope, flags breaking changes, requests changes before you look.
When a PR merges, Logger drafts a changelog entry and proposes a version bump (patch/minor/major).
If the PR changed public API, Docs updates the README + relevant /docs page in the same commit.
Weekly: Triager posts an issue digest โ open count by label, oldest unresolved, top requests.
You approve doc changes and version bumps in Telegram; everything else flows automatically.
Quarterly: a maintainer health check โ what's piling up, what could be deprecated, who's contributing.
Weekly maintainer digest (myrepo, 2026-04-29): - Triager: 14 new issues. 6 labeled 'bug', 4 'feature-request', 3 'docs', 1 spam closed. Oldest unresolved: #287 (32 days). Top request: TypeScript types (5 separate issues). - Critic: 8 PRs reviewed. 3 approved with comments, 4 changes-requested, 1 closed (out of scope). Avg review time: 12 minutes. - Logger: 5 changelog entries drafted. Proposed v2.4.0 (one minor breaking change in #312). - Docs: README updated for new `--watch` flag; API docs page regenerated; example snippet for migration from v2.3 โ v2.4 added. - Action items: review #312 breakage decision, decide on TypeScript types epic, acknowledge stale issue #287.
It catches the obvious things โ failing tests, style violations, scope creep, undocumented breaking changes. It will NOT catch subtle architectural problems or security issues that need deep context; the SOUL.md is calibrated to flag uncertainty rather than approve. In practice it raises the floor on what reaches your inbox: clean PRs get to you faster, sloppy ones get fixed before you see them.
Labels are easy to override โ Triager will defer to a human-applied label and won't re-label. The default config is conservative (it asks for clarification rather than guessing) so mislabels are rare. You can also tune Triager's SOUL.md with repo-specific rules: 'don't auto-close issues from contributor X', 'always tag accessibility issues as priority-high', etc.
Out of the box, no โ Critic only comments and requests changes. There's an opt-in `auto_merge` rule you can turn on in the SOUL.md (e.g. "auto-merge dependabot PRs that pass CI") but the default is human-merge only. We don't recommend auto-merge for non-trivial changes; the cost of a bad merge is much higher than the cost of a 1-day review delay.
On first run, Triager processes the backlog in batches (configurable, default 20 issues/run) to avoid overwhelming you with notifications. It dedupes against existing issues, so old duplicates get closed with a link to the canonical thread. Most maintainers see a 30-50% backlog reduction in the first week from dedup alone.
GitHub is first-class. GitLab and Gitea need a small adapter for the API differences (the bundle includes a `provider.ts` you swap). The SOUL.md prompts are platform-agnostic; only the API calls change.
Four SOUL.md files (Triager, Critic, Logger, Docs), an AGENTS.md coordination file, GitHub webhook + Actions workflows for the auto-triage pipeline, Dockerfile, and the setup README. All source is yours; runs on your machine, a $5/mo VPS, or as a GitHub Action.
Get 4 AI agents working together โ pre-configured, two Terminal commands to deploy.
7-day money-back guarantee ยท One-time payment, yours forever
AI DevOps automation team that runs CI/CD monitoring, PR review, and regression testing on autopilot for solo developers and small startup engineering teams.
Automated code review pipeline with 3 AI agents that scan every PR for security bugs, refactor opportunities, and regressions before merge - built for solo devs and shipping startups.
AI data analytics team that pulls Stripe/GA4/database metrics daily, flags anomalies before they become incidents, and benchmarks your numbers against industry priors โ built for SaaS founders without a data team.