docs-manager

Codex compatibility note:

• Invoke $skill-name: load the registered .agents/skills/<name>/SKILL.md and execute it with Codex tools; no separate Skill tool or Claude mode switch is needed. Author .claude/**; mirrors are generated.
• Track steps before execution and keep progress/evidence current. Explicit skill requests execute their protocol. Use the native question tool for required questions.
• Map foreign tool names to available host capabilities. A name mismatch needs no approval; an actually unavailable required capability stops with its step/evidence and a user question before adaptation.
• Subagent authorization: when a skill is user-invoked or AI-detected and its protocol requires subagents, that skill activation authorizes use of the required spawn_agent subagent(s) for that task.
• Preserve protocol order unless the user approves a deviation. Workflows use $start-workflow: gates stay fixed; other steps may flex with logged reasons and evidence. [BLOCKING] Mode routing — detect FIRST. Explicit --mode=update or --mode=init selects that mode. $docs-manager --mode=update and $docs-manager --mode=init are the former /docs-update and /docs-init: those slash commands no longer exist, and each mode works called directly with no workflow. No mode flag and no natural-language request matching the update Intent column: show the Mode Dispatch table and stop — ask nothing and run nothing (init is heavy and runs only on an explicit --mode=init). Read the mode file in full before anything else.

Quick Summary

Goal: Keep documentation true to code: update impacted context, config, specs/tests and derived/demo outputs; explicitly initialize or reconcile selected reference docs.

Summary: Update impacted docs or explicitly initialize reference docs: detect mode → read its contract → resolve scope/owners → run gates → review and report evidence.

Workflow: Detect mode → read its full reference → execute its contract → verify and report.

Key Rules:

• One mode per invocation; load only the selected mode's body.
• A natural-language request that matches the update Intent column selects update; init is selected only by an explicit --mode=init. No mode flag and no such request: show the Mode Dispatch table and stop.
• --mode= (two dashes) selects the skill mode. The update caller flag mode=update (no dashes, see that mode's Scope and inputs) only overrides $spec mode detection and never selects a skill mode.
• The docs-manager sub-agent (agent_type="docs-manager") is an agent, not this skill; it drives $docs-manager --mode=update.
• MUST ATTENTION keep claims evidence-based (file:line, confidence >80% to act) and task tracking live as each step starts and completes.
• Verify the shared output-quality contents check for each scoped document, including returned scan results; retain read-only audit boundaries.

Mode Dispatch

Mode	Purpose	Intent (natural-language triggers)	Read in full FIRST
--mode=update [modules=… changed_files=… phases=… skip_phases=… tc_mode=… freshness={impact|full|off} base=…]	Impact-scoped context/config checks → spec/test owners → derived/demo outputs → report. Formerly /docs-update	"update docs", "docs impacted by my changes", "sync docs after code change", "doc sync", "documentation update"	references/mode-update.md
--mode=init	Initialize or reconcile selected, applicable reference docs through scan. Formerly /docs-init	(explicit --mode=init only)	references/mode-init.md

• [BLOCKING] When --mode=update, read references/mode-update.md in full FIRST; it owns scope, owners, freshness, intent/coverage, caller flags and the report (tmp/reports/docs-update-{YYMMDD}-{HHMM}.md). Workflow invocation and standalone both run the full contract.
• [BLOCKING] When --mode=init, read references/mode-init.md in full FIRST; it owns optional config, document selection and applicability-gated scans.
• One-doc rebuilds stay in scan ($scan --target=<key>); refreshing every selected reference doc at once stays in scan-all. --mode=update escalates to them and --mode=init delegates to scan; neither is a mode of this skill.

Protocol guides — Apply each protocol only when its read-when condition holds. If a protocol's text is not in your context, read its file below before you act on it.

• output-quality-principles — Readable guidance and verified contents for long skill/doc files; writing, updating or auditing a skill or reference document, or another Markdown artifact over 200 lines → .claude/skills/shared/protocols/output-quality-principles.md

IMPORTANT MUST ATTENTION lead with useful guidance and readable priorities; preserve action-changing conditions/numbers and required structures. When writing, updating or auditing Markdown skill/doc content, verify top, heading-linked contents over 200 lines (stricter owner thresholds remain); refresh after edits and report gaps in read-only audits. Remove report bulk; judge semantic value rather than word counts.

Closing Reminders

IMPORTANT MUST ATTENTION Goal: Keep project documentation true to the code through the selected update or init contract.

IMPORTANT MUST ATTENTION Main steps: detect mode → read its full reference → resolve scope and ownership → run quality gates → review and report evidence.

• MANDATORY IMPORTANT MUST ATTENTION --mode=<x> reads references/mode-<x>.md in full FIRST; no mode flag and no update-intent request shows the mode table and stops — never guess a mode, never start init without an explicit --mode=init
• MANDATORY IMPORTANT MUST ATTENTION cite file:line evidence for every claim (confidence >80% to act)

[TASK-PLANNING] Track the selected contract before acting, with a final consistency review.

Evasion	Required action
"Direct call, skip the reference"	Standalone calls retain the full selected contract.