using-agent-skillslisted

# Using Agent Skills (router) Adapted from [addyosmani/agent-skills](https://github.com/addyosmani/agent-skills) `using-agent-skills` (MIT). Rewritten to route this harness's skill set. ## Two families of skills 1. **The gated pipeline** (feature delivery) — Spec Kit stages plus this harness's stages, with human gates. Use `/pipeline` to find the current stage. Order: `/task <issue>` (intake + branch) → `/grill-with-docs` (**mandatory** alignment) → `/speckit-specify` → `/speckit-clarify` → `/speckit-checklist` → `/speckit-plan` → `/speckit-tasks` → `/speckit-analyze` → **gate 1** → *(optional: `/gherkin` → gate 2)* → `/implement` → `/refactor` → `/architect` → **gate 3** → `/ship-review` → rebase → PR. Acceptance tests (Gherkin/Reqnroll) are **opt-in** — skip the Gherkin stage and gate 2 unless requested. 2. **Supporting skills** (non-gated) — pulled in as needed during the pipeline. This router maps tasks to them. ## Discovery decision tree ``` Task arrives ├─ Where am I in the pipeline? ─────────────→ /pipeline ├─ Starting ANY task (have an issue) ───────→ /task <issue> (fetch issue + branch) ├─ Branch created, before spec/code ────────→ /grill-with-docs (MANDATORY alignment) ├─ Alignment done → write the spec ─────────→ /speckit-specify ├─ Spec has ambiguities ────────────────────→ /speckit-clarify ├─ Need plan / tasks / consistency check ───→ /speckit-plan · /speckit-tasks · /speckit-analyze ├─ Acceptance scenarios (OPTIONAL, opt-in) ─→ /gherkin (only i