← RCF Lite docs

RCF Lite

For agents

Agent-consumable content, single-sourced from the shipped rcf-lite guidance pack. If you supervise an agent that runs the method, read here to see the rules it has been told.

Every page in this section is a projection of a shipped file. The docs build pulls the pack at build time from rcf-lite@0.15.0 installed in the site's own node_modules; a drift test in CI refuses the build if the shipped bytes have moved since the site's checked-in copy. When your agent has an MCP connection, it reads exactly the same content; when it does not, these public URLs are the fallback.

  1. Harness block

    The managed block rcf init writes into CLAUDE.md and AGENTS.md. The operating rules your agent follows, verbatim.

  2. Elicitation playbook

    The playbook the agent walks with you before it writes a document. What it will ask, and how the answers become the chain.

  3. Build-cycle playbook

    The five-stage loop the agent runs for every build item: define, build, review, test, finalise. Each stage commits.

  4. Document model

    The types the chain uses, their fields and relationships. The spec the agent authors against.

  5. Persona programme

    The tail-interview template for programme-scale work. Optional; used when the elicitation needs it.

  6. MCP surface

    The tools and prompts the rcf MCP server exposes to a connected agent, one entry each.

New here as a supervising human? Start with the harness block. It is the rules your agent is being held to, in the language the agent reads.