What is openfabriq
A software factory described in readable files, run by an engine on your computer or your own server.
openfabriq runs a software factory: a small team of AI agents that triage issues, implement them, review pull requests, investigate errors or keep documentation up to date. Each agent works at a moment you choose.
You describe the factory in an openfabriq/ folder of your repository, in files that a non-developer can read. The engine reads that folder, watches GitHub, starts the right agent with your Claude Code subscription or an Anthropic API key, and writes back to GitHub only what you allowed.
The model is the one of Terraform. You describe the factory, openfabriq plan shows what will change, and openfabriq apply applies it.
How it fits together
Your repository holds the description of the factory, and one machine runs it.
- The openfabriq/ folder: agents, automations, skills
- Versioned and reviewed like the rest of your code
- Started with openfabriq start
- Runs the agents with your Claude Code
- Serves the studio, a local app to follow every run
- Issues, pull requests and reviews
- Slack, Linear and Sentry, if you connect them
The engine is the same program on your computer and on a server: start locally, and move to a server when your team needs it to run at all hours.
A factory is a folder
A complete one, file by file:
One file says where the factory works, and what it may spend.
The repository, the people whose actions start it, the checks that prove a change, the services it listens to, and the spending limits.
name: Web app factory
description: Agents that keep the web app moving, from issue to pull request. A person reviews and merges every change.
repository: acme/web-app
maintainers: [alice] # only their actions start the factory
model: sonnet # the default model of the agents
setup: pnpm install --frozen-lockfile # prepare each run's checkout
commands: # the project's commands, by name
test: pnpm test
lint: pnpm lint
checks: [test, lint] # run by the engine before any pull request leaves the machine
integrations:
sentry: # new production errors of this project are fixed
organization: acme
project: web-app
webhooks:
deploy-failed: # the deploy platform posts here, through the relay
signature: standard
mcp:
docs: # up-to-date documentation of the libraries the code uses
url: https://mcp.context7.com/mcp
limits:
usd_per_run: 5
usd_per_day: 30
answer_mentions: true # the agents that open pull requests answer a maintainer's @openfabriq in a comment on them; false turns it offThe real files of examples/00-at-a-glance. Select a file to read it.
A maintainer adds the ready-for-agent label to an issue. Within a minute, the implementer starts in a fresh checkout. A few minutes later, a pull request is open, and it has passed the checks declared in factory.yaml. A new production error in Sentry follows the same path on its own, and a failed deploy becomes an investigated issue. A person merges.
Principles
- Agents read, the engine writes. An agent has no GitHub credentials. It proposes outputs, and the engine applies only those that
canallows. A person merges. - What runs is what was applied. The engine runs the copy made by
apply, never your branch: a pull request cannot change the rules that judge it. - Readable by anyone. Triggers (
when) and rights (can) are words from a closed list, checked byopenfabriq validate. - A run never waits for a person. It ends. People act on GitHub with a label, a comment or a merge, and that starts the next run.
- Nothing is exposed. The engine uses your Claude Code subscription (or an API key on a server) and your
ghlogin, and opens no port to the Internet.
Why openfabriq explains each choice, and Concepts defines each term. To try it, install openfabriq, then follow the Quickstart.