docslisted

Run an interactive Docs authoring session for a single 📝 Docs task: load the task, read its declared Source domain(s) live over the web, author into its declared Target surface (a repo file or a Notion page), and open the never-auto-merge output (a draft PR for a repo-file target, or a staged notion.pageEdit intent for a Notion-page target) for a human to review and merge/apply. Use when the user says "docs", "docs session", "run the docs task", "write the docs for X", or starts an interactive Docs-authoring session on a 📝 Docs task. Requires the task to declare a Target surface and Source domain(s) in its body (see § Docs task-body convention below) — do not improvise either.
phahadek/claude-orchestrator · ★ 1 · Web & Frontend · score 74

Install: claude install-skill phahadek/claude-orchestrator

# docs — author a single Docs task Companion to `/groom` (backlog→Ready) and `/design` (locks the spec a Docs task follows): this skill *executes* one 📝 **Docs** task — an authoring pass against a live external source, landed as a never-auto-merge output for a human to apply. It is dispatched under the `docs` session type, whose tool set (`DOCS_ALLOWED_TOOLS`) is the first session profile in this repo with web egress: `WebFetch`, allowlisted **only** to the task's declared Source domain(s) — never open `WebSearch`, never an un-allowlisted fetch. This is the same authoring loop whether a human triggers it interactively (this skill) or a future autonomous trigger dispatches it — only the trigger differs. The interactive core built here is what that autonomous wiring will reuse. ## Docs task-body convention A 📝 Docs task's body must declare, before this skill starts writing anything: - **Target surface** — where the authored content lands. Either: - a **repo path** (e.g. `docs/api/webhooks.md`) — output is a draft PR against that repo,  opened by this session (`Bash(gh pr create:*)` / `mcp__github__create_pull_request` — see § Output below for why this session, uniquely, carries that capability); or - a **Notion page id** — output is a staged `notion.pageEdit` intent (find/replace pairs against the page's current body), left for an operator to apply. - **Source domains** — the domain(s) this task's `WebFetch` allowlist is scoped t