harness-loom

Produces the goal markdown that `/harness-orchestrate` consumes as the original user request. Use when `/harness-goal-interview` is invoked, when the user is starting a harness cycle on a new feature or initiative and needs to articulate requirements precisely (e.g. 'write down what we're trying to do' before running the harness), or when an earlier `/harness-orchestrate` invocation failed because the goal file was vague or plan-shaped. Runs a full user interview focused on requirements precision and user-side technical decisions — implementation detail is the planner's and producer pairs' job, not the goal file's — then writes the five canonical sections (Why / Goal / Constraints / Out of scope / Completion criteria) to the target project root. Apply this skill whenever the user is about to hand a goal file to `/harness-orchestrate` even without an explicit interview ask; a ten-minute structured elicitation pays back across every EPIC, task, and review the cycle produces.

Use when `/harness-init` is invoked to install the target-side runtime foundation into the current working directory. Scaffolds `.harness/loom/` canonical staging plus `.harness/cycle/` runtime state, copies target-local `hook.sh` and `sync.ts`, and stops there. Platform trees are derived later by `node .harness/loom/sync.ts --provider <list>`.

Use when `/harness-auto-setup` is invoked to bootstrap and configure a project-shaped harness, expand an existing harness without refreshing its foundation, or migrate an existing harness foundation with snapshot-first preservation.

Use when `/harness-pair-dev` is invoked to author, revise, position, or remove a target project's producer-reviewer pair in `.harness/loom/`.