inferspec-caplisted

# /inferspec-cap Take ONE capability deep — solicit Jira/Confluence/URL context the user has, draft (or reload) its OpenSpec `spec.md`, then ask the user one focused question per `[GAP]` marker until the spec is converged. `/inferspec-scan` writes ambient drafts for the whole repo. `/inferspec-cap` is how you turn one draft (or no draft) into a firm spec. ## Output Writes / overwrites `openspec/specs/<slug>/spec.md` in OpenSpec format. Adds `last_qa_run` field to the `__inferspec_meta__` footer. On exit (with user's consent), commits the change. ## Usage ``` /inferspec-cap <slug> # explicit slug from features.json /inferspec-cap "user auth" # fuzzy match /inferspec-cap # interactive picker /inferspec-cap <slug> --new # bootstrap a new cap /inferspec-cap <slug> --save-transcript # save Q&A log ``` ## When this skill is invoked, run these steps ### Step 0 — Parse flags + load context Parse: - `--new` → bootstrap mode - `--save-transcript` → write `openspec/specs/<slug>/_qa_log.md` (gitignored) Read `.inferspec.yaml` for any `mcp_overrides`. ### Step 1 — Resolve the cap Read `prompts/resolve_cap.md` and follow it. Get a resolved `cap = {name, files, existing_spec, existing_markers}`. If `features.json` is missing, offer to run `/inferspec-scan` first. If user agrees, invoke it then restart this step. ### Step 2 — Solicit external sources Read `prompts/solicit_sources.md` and follow it. Gather `cap.external_context