vladmesh

Scan codebase for dead code, code smells, security issues, contract violations, missing DTOs, and architectural deviations. Creates/updates docs/audit.md and adds actionable items to backlog. Use when user says "audit", "scan", "check code quality", or wants to find hardcoded strings, bypassed abstractions, missing schemas, untyped API boundaries, or drift from shared contracts.

Structured thinking session on a topic. Creates docs/brainstorms/<topic>.md with Status tracking and Action Items. User routes results manually (new sprint, backlog, VISION update, hotfix).

Close the current sprint phase — verify all tasks done, run/write integration tests, advance to next phase.

Close the current sprint — final gate after endgame. Push all commits, update STATUS.md history, update CHANGELOG.

Sprint dispatcher — reads STATUS.md and invokes the right skill based on current sprint state. Use when user says "go", "next", "continue", or wants to proceed with sprint work.

Implement the current sprint task using TDD. Reads task from sprint directory, creates git branch, updates task status on completion. Main development skill.

Create a new sprint — checks tech sprint cadence, proposes scope from VISION/ROADMAP/backlog, creates sprint directory and files.

Generate task files for the current sprint phase. Reads sprint.md, creates 2-4 task files with descriptions, tests, and acceptance criteria.

Update living documentation to match current codebase state. Two modes: `/update-docs` — incremental, checks git log since last run; `/update-docs full` — deep-reads actual code and compares with all docs. Use when user says "update docs", "docs are stale", "sync documentation", or after a batch of changes lands.

Advance the dnd-simulator sprint pipeline by reading repository state, selecting the next project playbook, and executing exactly one pipeline step. Use when the user invokes /go or $go, or says go, next, continue, дальше, продолжай, or asks Codex to keep the current dnd-simulator sprint moving.

Close the current sprint: verify all phases done, audit completed, post-audit E2E green, integration tests green, then update docs, commit, push, and wait for CI. Blocks if anything is off. Use when user says "close sprint", "finish sprint", "sprint done", "wrap up sprint", or when all phases are complete and it's time to finalize.

Determine the next step in the sprint pipeline and execute it. Reads docs/STATUS.md, analyzes current state, and calls the appropriate skill. Use when user says "go", "next", "continue", "what's next", "продолжай", "дальше", or just wants to keep the pipeline moving without thinking about which skill to invoke.

Run E2E regression tests via Playwright against the live app. Follows the playbook in docs/e2e-playbook.md plus auto-discovered scenarios from recent changes. Writes a report to docs/e2e-reports/. Use when user says "e2e", "run e2e", "regression", "test everything", "full test", "smoke test", or wants to validate the app works end-to-end. By default skips LLM scenarios; pass --with-llm to include them. Can also be invoked by other skills (close-phase, close-sprint) as part of their workflow.

Implement the next pending task in the current sprint phase. Follows TDD: write tests first (RED), then implement (GREEN), verify all tests pass, update docs, commit. Use when user says "implement", "next task", "work on task", "start task", "do the task", or when the user wants to pick up the next piece of work in an active sprint. Also trigger when user says "implement task N" to target a specific task.

Plan a new sprint: choose scope from backlog/roadmap, break into vertical phases, create sprint docs. Use when user says "new sprint", "next sprint", "plan sprint", "start sprint", "sprint planning", or after a sprint is completed and it's time to pick up the next chunk of work. Also trigger when user asks "what should we work on next" or "what's the priority now" in the context of this project.

Fully autonomous sprint orchestrator — an experimental hands-free development format with NO human in the loop. Runs the full sprint pipeline by spawning one fresh agent per pipeline step; each agent does exactly one granule (/plan-phase, /implement one task, /close-phase, /audit, etc.) and exits. The orchestrator is the SOLE decision authority: it makes every call itself, answers any clarifying questions its sub-agents raise, and ends the run only on success or an unrecoverable dead-end — never to ask a human. Optionally bounded: `/meta-go N` runs at most N granules then stops with a resume report. Use when the user says "meta-go", "run the sprint", "autopilot", "автопилот", "прогони спринт", or wants to leave and come back to a finished sprint.

Процедура куратора памяти — из свежих транскриптов вытащить durable-факты, дедупнуть, записать их через secretary memory protocol. Запускается автоматизацией Orca (Claude/Sonnet) в воркспейсе ~/triggered-agents. Куратор — первый плагин рантайма triggered-agents.

Процедура ретро-агента — пройти по свежим транскриптам голов и memory-mcp search-log, найти конкретные фейлы (ответ по факту из panelmem без memory_search и с ошибкой, повтор известной ошибки, зацикливание, пустая сессия) и оформить ПРЕДЛОЖЕНИЯ карточками в колонку Идеи на доске Pipeline. Сам ничего не внедряет, в Ready карточки не двигает. Третий плагин рантайма triggered-agents, запускается автоматизацией Orca daily в воркспейсе ~/triggered-agents.

Процедура агента-стюарда — часовой присмотр за пайплайном плюс раз в сутки безусловный обход всей системы (без precheck-гейта, ловит в том числе слепоту самих сигналов). Основная цель — живой пост-мерж контроль самомодификации: e2e для мета-проектов (сама доска/пайплайн/куратор) невозможен иначе, чем на живой системе, поэтому стюард просыпается по сигналу аномалии или по суточному расписанию, разбирается по данным системы, чинит руками то, что блокирует пайплайн прямо сейчас, заводит карточки на остальное. Плюс рутина: триаж Blocked, стойла, сироты-воркспейсы, аномалии в логах. Запускается автоматизацией Orca (голова claude-fable) в воркспейсе ~/triggered-agents.