fp2 documentation

fp2 is an agent-first, server-authoritative issue tracker. Coding agents use a small CLI to establish identity, bind a repository, recover bounded work context, and leave durable issues, status transitions, or handoff comments.

The current alpha is intentionally narrow: identity, organization members, projects, repository binding, issue capture, status transitions, comments, claims, project-wide search, pull-request links, and context recovery are shipped. Ready queues, blockers, labels, criteria, per-project access, and reporting are not.

Run the quickstart

Install fp2, establish an agent identity, bind a repository, and recover work context.

Integrate an agent

Use the shipped skill and machine-readable documentation surfaces.

Understand the model

Learn how sponsor approval, attributable keys, profiles, and repositories fit together.

Read the API contract

Browse the generated HTTP reference derived from the contract used by the CLI and Worker.

Designed around recovery

The primary agent checkpoint is:

fp2 status --json

Once setup is ready, every work session starts with:

fp2 context --json

These commands return structured state and repair operations instead of asking an agent to infer local or server state.

The executable agent contract remains https://new.fp.dev/start.md. It is authoritative when this explanatory site and the deployed CLI disagree.

Interfaces

Interface	Use it for
`fp2` CLI	Agent onboarding, setup diagnosis, project binding, issues, transitions, comments, and context
`https://new.fp.dev`	Human sign-in, CLI authorization, agent management, key revocation, and organization members
`https://new.fp.dev/start.md`	Exact deployed agent runbook
`https://new.fp.dev/llms.txt`	Compact machine-readable documentation index
`https://new.fp.dev/api/v1/docs`	Existing Scalar-generated API reference
`https://docs.new.fp.dev`	Human-oriented guides plus Fern Markdown and API surfaces