Introduction

CAF (Coderium Agent Framework) runs AI agents as an engineering team that takes a ticket from spec to pull request, with strict governance along the way.

Every task runs through the Plan → Implement → Verify → PR cycle, and starts from a written spec: the Planner writes it, a human can approve it, and every later agent is checked against it (see Spec-Driven Development). Each gate that fails stops the pipeline and hands the work to a human, and nothing is ever merged automatically.

Two core components

• CAF Initiator — the caf-init CLI. It detects your repo's stack, package manager and ticket tracker, then writes CAF's starter files: knowledge base, agent definitions, workflow docs, slash commands and skills. It never calls an AI model; anything it can't detect is left as a TODO.
• CAF Orchestrator — a webhook receiver (Fastify + BullMQ + Redis) that runs on your own VPS. When a ticket becomes ready — a Linear workflow state or a GitHub Issue label — it runs the agents in sequence and opens the pull request. It asks on the ticket when the spec has an open question, and can wait for a human to approve the spec before any code is written. Jira support is planned but not yet implemented.

You can use CAF Initiator on its own: the agents it generates also run directly in Claude Code, without the Orchestrator.

The agents

CAF ships five agents: caf-planner, caf-frontend, caf-backend, caf-validator and caf-reviewer. See Layer 2: Agent Definitions.

Continue to Quick Start to set up CAF in your repo.