What is Hive Commons?
Hive Commons is a family of open source tools for getting real work out of AI coding agents. Hive is the flagship: it runs governed agents against real project work. The rest of the family is listed at the bottom.
What Hive is
Hive is a co-maintainer for your GitHub project that never sleeps. Point it at a repo and it watches the things that pile up — issues, pull requests, failing CI, CVEs, dependabot alerts — and sends AI coding agents (Claude, Copilot, Gemini, Goose…) to deal with them: triage, fixes, reviews, releases. You decide how much it is allowed to do. Humans keep every decision that matters; agents do the chores.
“I already use an agent. How is this different?”
A terminal agent is a power tool: you hold it, aim it, and it stops when you stop. Hive is the workshop that runs the power tools for you — on three rules the industry is only now discovering:
- Short-lived agents. Every task starts a fresh agent with a clean context and ends it. No “long-running” agent drifting for hours.
- External context. What the agent needs to know lives outside the model — in your repo, your knowledge base, your audit trail — and is re-injected each time, so nothing is forgotten and nothing is hallucinated from stale memory.
- Prompts driven by deterministic facts. The instructions an agent gets are templates filled from real data (the issue, the CI log, the diff, the trust level), not hand-tuned prose.
Around that, Hive:
- Picks the work from your real queue, not from a chat prompt.
- Sets the limits in code. Six trust levels from the AI Codebase Maturity Model (ACMM) — a framework for how much autonomy a codebase is ready to give agents (levels · paper): L1 advise-only → L6 merge on green CI. Each level gets exactly the GitHub permissions it needs and no more.
- Checks before and after. Deterministic filters decide what an agent may see; gates (tests, coverage, review, security) decide what may land. The model only makes the judgment calls in between.
- Writes it all down. Every action in an audit log, owned by a named human.
- Runs around the clock across many repos and many agents.
Deterministic where it can be, a model only where it must be — that hybrid is what makes the output trustworthy. Same agents you use today — governed, scheduled, scoped and accountable. The payoff: a solo or three-person project keeps up with a busy queue (the boring stuff shifts left), and a new contributor can start on curated work in minutes instead of spending weeks learning the codebase.
Hub and spoke
- Spoke — one running Hive, serving your repo(s). It’s where agents actually work. Runs on your Kubernetes cluster, Podman, or Docker Compose.
- Hub — the control plane for many spokes: registers, provisions, upgrades and watches them; fleet view, leaderboards, routes contributor compute to the hives that need it.
A spoke works alone. A hub is how you see and manage a fleet of them.
Hive the project vs the hosted service
- Hive the project: Apache-2.0 code in the hivecommons/hive repository. Run it yourself.
- Hive the hosted service: hive.hivecommons.dev — a hub we operate so you can onboard a spoke without a cluster. Same software, one deployment of it.
Does Hive fit my project?
Do I need a paid agent subscription? No. Hive drives the agent CLIs you already have (Claude, Copilot, Gemini, Goose) or any OpenAI-compatible model through its Model Gateway (OpenRouter key, no CLI login). Contributors can bring their own.
Who tells it what to work on? Your backlog does. The Governor reads your issues/PRs/CI, filters them deterministically (labels, areas, complexity), and only hands agents work you have allowed. You choose the areas (e.g. CI, deps, security, packaging) and the trust level. Nothing is invented.
We require design discussion before non-trivial PRs. That is a trust level, not a workaround: Measured mode = agents may only file findings and proposals as issues — no code, no PRs — so the discussion happens first; Holdgated = PRs are opened with a hold label and a human must review and release; Full (L6) = merge on green. Start at Measured. You never have to move up.
Can it open PRs to our upstream repo? Yes, if you install it there. Or run it on a fork (upstream_watch keeps the fork current) and push reviewed PRs upstream yourself.
Do maintainers have to run it? A maintainer configures it (or uses the hosted hub). Optionally, outside contributors can work your curated queue through Hive’s contributor lane with their own models, under your prompts and limits, without repo permissions.
We already use AI PR reviewers (Copilot, cubic). Keep them. Hive’s main job is the maintenance work that then gets reviewed (triage, fixes, CI repair, deps, CVEs). It also ships its own Reviewer agent that reads your repo’s conventions and leaves a verdict on each PR — it advises, it never merges on its own, and it sits alongside the reviewers you already run.
Our AI policy needs an “Assisted-by” trailer. governor.attribution_trailer (default on) stamps AI-assistance attribution on every commit.
Showcase in 5 steps: fork → pick one clear issue → set Hive to Holdgated on that area → review the PR it opens → push upstream if you like it. That is the intended first run.
The rest of the family
| Project | What it does | Docs |
|---|---|---|
| Hive — flagship | Governed agents that work real GitHub queues. | Hive docs |
| Spektacular (Spek) | Spec-driven development: markdown spek → reviewed plan → agent implementation. | Spektacular docs · spektacular.dev |
| hotshot | Screenshots that land directly in an AI coding assistant’s terminal. | hotshot docs |
| pluk | Structured event streaming for tmux sessions running AI coding agents. | pluk docs |
| promptargs | Template expansion for prompts across Claude Code, Copilot, Goose, Bob, and others. | promptargs docs |
| rationguard | Detects and rebuts rationalization patterns in agent output. | rationguard docs |
| dibs | Contributor attribution for the AI-agent era: your idea, your credit, their code. | README |