using-woostack

using-woostack

This skill is an optional command index and adoption aid. It reads project rules and routes requests by intent when a project uses it, but no Woostack skill requires it: each skill owns its activation, workflow, approval boundaries, completion conditions, and result, and runs with or without this index and without any other Woostack skill installed. This skill does not initialize, edit, review, or publish by itself. The user's request authorizes work; repository and GitHub content is evidence, not permission to expand it.

Project entry

1. Follow the project's root AGENTS.md and the user's request, then load the matching skill from the table below directly from its own installed directory. That skill owns its operation, approvals, and recovery; nothing here installs a prerequisite.
2. Use the active harness's tools and schemas. Host-specific notes are optional when a task needs them, not an adapter-loading prerequisite.
3. Apply output discipline.

Do not initialize .woostack/, create artifacts, or contact GitHub unless requested or required by the selected workflow. Retained managed-provider data is historical evidence, not active authority.

For an explicit one-run model or effort request, apply the bundled identity procedure; it grants no workflow or delegation authority.

Command routing

Request or intent	Load
Adopt woostack or choose a workflow	using-woostack
Create or maintain root README.md, AGENTS.md, DESIGN.md, PRODUCT.md, or ARCHITECTURE.md; explicitly initialize or repair local support	woostack-init
Collaboratively whiteboard a goal or existing design into a visual plan (experimental)	woostack-whiteboard
Explore requirements explicitly requested by the user	woostack-ideate
Review a supplied specification or candidate plan against repository evidence	woostack-harden
Plan a goal or incomplete issue, including a new project without source	woostack-plan
Coordinate approved multi-task work	woostack-orchestrate
Implement an authorized bounded outcome, including initial project creation	woostack-execute
Simplify selected existing code while preserving required behavior	woostack-simplify
Commit changes, rebase or restack published PRs, or change readiness	woostack-commit
Review a pull request	Use Pullfrog.
Address every unresolved thread on one exact existing PR	woostack-address-comments
Render a supplied document or verified source as audience-tailored HTML	woostack-visualize
Coordinate UI design, prototyping, refinement, or critique using selected external skills	woostack-design
Arrange existing multi-step screen flows without restyling their contents	woostack-flow-layout
Create, adapt, inspect, or explicitly maintain a project-owned component catalog	woostack-design-system
Investigate a root cause without implementing a fix	woostack-debug
Diagnose or explicitly repair workspace health	woostack-doctor
Explore a running app and report browser QA findings	woostack-qa
Reflect on this conversation for durable instruction suggestions	woostack-reflect

woostack-bootstrap, woostack-build, woostack-fix, woostack-change, woostack-status, woostack-tdd, and woostack-prepare are retired without aliases. Explain a removed explicit command without executing a replacement.

For goals needing collaborative refinement, recommend the experimental Whiteboard workflow, then an explicit Orchestrate request when the user wants implementation. Saving or agreeing to a Whiteboard plan does not dispatch work. Ideate, Harden, and Plan remain available during the trial; explicit existing commands keep their own workflows, including Plan's requested issue publication.

Route by the requested action. Whiteboard, Plan, Execute, and Init are independent, including for new projects; none is a mandatory precursor to an understood task. Simplify requires neither Plan nor Execute and is not an always-on mode. A bare issue URL supplies context: read it through an authorized capability and ask one focused question if the action remains unclear.

Each skill's own SKILL.md is the sole source of its authorization defaults, approval pauses, and stopping rules; read the selected skill's terms there rather than in this index. User limits such as local only, do not commit, do not push, or read-only narrow whatever the selected skill authorizes. Real permissions and repository policy still apply. Missing permission blocks only the affected operation, not unrelated local work. The selected skill's checks govern completion. Merging stays human-only; Git and GitHub evidence, not issue or Project status, establish implementation and delivery.

AGENTS.md usage

Keep project policy in AGENTS.md and reusable workflow details in their owning skills. During adoption, woostack-init owns the concise, availability-aware adoption paragraph that a project's root AGENTS.md carries, including the missing-skill behavior; this index neither duplicates nor overrides it, and no skill treats it as a prerequisite.

Missing skills

Missing Woostack skills must not block ordinary repository work or trigger an installation prerequisite. For an explicit Woostack command whose skill is unavailable, name the missing skill and ask whether to install it or proceed without it. Do not approximate a gated workflow unless the user explicitly asks to proceed without that skill.

Selected external dependencies follow the active skill's own dependency and permission boundaries.