knot

Create looms, knots, and profiles by writing .md files directly. Knot auto-discovers looms (directories ending in `-loom`) and parses knot definition files (`.md` files inside loom directories). Profiles live in `rig/profiles/`. Read `rig/state.json` to verify state after file changes. USE FOR: create loom, add loom, new loom, delete loom, remove loom, modify loom, update loom, create knot, add knot, configure knot, loom CRUD, knot CRUD, loom management, knot management, create profile, agent profile, profile CRUD. DO NOT USE FOR: initialising a rig (use knot-init), inspecting state (use knot-inspect), triggering processing, running agent sessions.

Initialise a Knot rig in the current directory. Detects if a rig exists, verifies Knot is running by checking `rig/state.json`, and creates the rig directory structure. If no profiles exist, creates a default profile by reading available models from ~/.pi/agent/models.json. Verifies setup by reading `rig/state.json`. USE FOR: init knot, knot init, setup knot, configure knot rig, start knot, initialise knot, knot configuration, rig init, rig setup. DO NOT USE FOR: creating looms, creating knots, inspecting loom state, modifying existing looms.

Design looms and knots for the Knot agent orchestration framework. Covers idempotency, naming conventions, responsibility boundaries, domain direction, loop design, and loop-breaking patterns. USE FOR: design knot, design loom, knot design, loom design, knot architecture, agent loop, feedback loop, knot naming, strand direction, knot responsibility, idempotent knot, loop-breaking, knot workflow design. DO NOT USE FOR: creating looms/knots (use knot-create), initialising a rig (use knot-init), inspecting state (use knot-inspect).

Inspect the current state of a Knot rig: list looms, examine loom details, view activity logs, check knot processing status, list agent profiles. Read rig state from `rig/state.json` and activity from `rig/tie-offs/{loom-id}/.loom-log`. USE FOR: inspect rig, check rig status, view looms, list looms, inspect loom, loom status, knot status, check knot, view activity, loom activity, processing status, knot state, rig state, what looms exist, show looms, loom details, list profiles, view profile, check profile. DO NOT USE FOR: creating looms (use knot-create), deleting looms (use knot-create), creating profiles (use knot-create), initialising a rig (use knot-init), triggering processing.

Record format changes between Knot binary versions. When a project updates its Knot binary, this skill tells the agent what changed in project documents (profiles, knots, looms) and how to migrate them. Contains a versioned changelog with migration instructions for each breaking change. USE FOR: update knot, knot version change, knot migration, knot changelog, migrate knot documents, knot format change, knot upgrade, knot breaking change, profile format change, knot file migration, loom migration. DO NOT USE FOR: creating looms (use knot-create), modifying looms (use knot-create), initialising a rig (use knot-init), inspecting state (use knot-inspect), fixing bugs (use project-bugfix).

Generate user-facing documentation from project artifacts (skills, glossary, PRDs, plans, API code). Extracts configuration reference, concept guides, workflow tutorials, API reference, and release notes from internal project documents and reformats them for human readers. USE FOR: generate docs, write documentation, user docs, doc generation, extract documentation, documentation from skills, docs from PRD, docs from plans, API reference docs, release notes, changelog generation, documentation methodology, doc extraction, write user guide, create documentation, documentation skill. DO NOT USE FOR: writing ADRs (use architecture-decision-record), writing DPRs (use design-pattern-record), creating plans (use project-planner), implementing features, writing code, running tests.