andrei-shtanakov

Guide a new contributor into the AI-Orchestrators ecosystem — walk a role checklist step by step, answer every question grounded in the KB with a citation, mark progress in the runtime store, and capture questions the KB cannot answer as KB-gap signal. Robin's first paid duty (ROBIN-SPEC §6.3; robin/duties.md

Curator of the Ecosystem KB (prograph-vault). Keeps the knowledge base in order: freshness audit of the registry and authored docs, archiving of the obsolete (with date+reason, no deletion), fixing broken links, knowledge search, graduation of canon from _cowork_output into the KB, and pruning/archiving the derived/journal. Run when the user asks: "tidy up the KB / knowledge base", "check the registry's freshness", "archive the obsolete", "fix the links in the vault", "find … in the KB", "move this from _cowork_output into the KB", "clean up authored", as well as on a scheduled (e.g. weekly) run. Works ONLY in the KB; other repos are read-only. Filesystem-only, no Obsidian.

Proactively warm up context from the Ecosystem KB (prograph-vault) when starting work on a known area — interfaces, cross-cutting rules, contracts, or established patterns for the current project. Loads the relevant bundle at once (applicable rules + this project's facts/interfaces + related contracts + recent journal). Use at the start of a work area; for a single targeted question use kb-search instead.

Record a significant action on the CURRENT project into the Ecosystem KB (prograph-vault) journal. Use after any notable project event — a decision, an interface/contract change, a new component, a migration, a status change, or a noteworthy result. Not just hard problems: log all significant project actions. Scoped to the current project only; appends to derived/journal/<project>/journal.md. Does not auto-commit.

Answer a targeted question against the Ecosystem KB (prograph-vault) — standards, cross-cutting rules, templates, contracts, project facts, recent changes, or past activity. Use for ad-hoc lookups like "what's the rule for X", "which template for Y", "what changed in Z lately", "did we decide about W". Read-only, unrestricted across the whole KB. For proactively loading a work area up front, use kb-load instead.

Orient the session in the Ecosystem KB (prograph-vault) at start. Called automatically via a SessionStart hook. Read-only — detects the current project, surfaces applicable cross-cutting rules and the latest project journal entries, and points to /kb-load, /kb-save, /kb-search. Never writes to the KB.

Create project specifications in Kiro-style format — three linked documents (requirements, design, tasks) with requirements-to-tasks traceability. Use when starting a new project spec, documenting an existing project, or on requests like "create a spec", "write requirements", "document the project".

Generate game theory test scenarios — YAML game suites and pytest tests for game-environments (8 games, 25+ strategies) and atp-games. Use when asked to create game tests, game scenarios, game suites, test game strategies, or evaluate game-theoretic properties. Triggers on "game test", "game scenario", "game suite", "test game", "тесты для игры", "игровой сценарий", "тестовый сценарий для теории игр", "prisoners dilemma test", "auction test", "blotto test", "stag hunt test", "battle of sexes test", "el farol test".

Generate pytest tests for ATP platform modules. Use when asked to create tests, write tests, or generate test coverage for any ATP component (evaluators, adapters, runner, protocol, loader, etc.). Triggers on "generate tests", "write tests for", "create tests", "add tests", "cover with tests", "напиши тесты", "сгенерируй тесты".

Generate game theory test scenarios — YAML game suites and pytest tests for game-environments and atp-games. Use when asked to create game tests, game scenarios, game suites, test game strategies, or evaluate game-theoretic properties. Triggers on "game test", "game scenario", "game suite", "test game", "тесты для игры", "игровой сценарий", "тестовый сценарий для теории игр", "prisoners dilemma test", "auction test", "blotto test".

Generate pytest tests for ATP platform modules. Use when asked to create tests, write tests, or generate test coverage for any ATP component (evaluators, adapters, runner, protocol, loader, etc.). Triggers on "generate tests", "write tests for", "create tests", "add tests", "cover with tests", "напиши тесты", "сгенерируй тесты".

Generate a product-knowledge layer (a "brain") for a team's codebase - glossary and object model mined from real code, rot-rate tiers, a claims-manifest drift checker that fails commits when docs disagree with code, an eval harness, and a weekly sync skill. Use when the user says "init team brain", "create a product brain", "set up drift-checked product docs", or wants onsa-brain-style product knowledge for their repos.

Mine how YOUR team actually decides into a decisions.md — dated, sourced decision principles with statuses, scopes, and one-way doors — and generate a decide-like-<your-team> apprentice-loop skill: teammate commits to a recommendation first, the skill cites principles and gap-analyzes the reasoning, episodes get logged for calibration, and consistently-calibrated teammates graduate to owning whole decision classes. Use when the user says "init team decisions", "set up a decision system", "capture how we decide", "build a decide-like-us skill", or keeps getting pinged with "what would <the founder> say?".

Evidence-first Team OS setup. Mines the host agent's local transcripts (Claude Code, Codex, …), MCP config, and existing repos to pre-populate a team knowledge repo interview, asks at most 5 questions, builds the minimal repo that answers the team's top recurring question, and ends with a ready-to-post demonstration message. Use when the user says "init team os", "set up our team os from evidence", "build our team knowledge repo", or wants a TeamOS scaffolded from what their machine already knows.

Configure a personalized Robin (AI chief of staff) spec for any team. Reads the user's TeamOS knowledge repo, mines MCP config and installed runtimes to resolve ROBIN-SPEC's implementation-defined slots, drafts the team's soul.md and duty roster into their repo, and emits ROBIN-SPEC.local.md plus a KICKOFF.md prompt for any coding agent to implement Robin on the user's stack. Use when the user says "robin init", "configure my robin spec", "set up a team agent spec", "set up a chief of staff agent", or wants to start implementing a Robin-style team agent.

Implement a spec-runner task end-to-end with the pi coding agent — write the code AND its tests, run the tests via bash until green, then emit the spec-runner completion marker. Use during the development stage of the dev→review→test loop.

Read-only code review for a spec-runner task with the pi coding agent — inspect the diff and changed files, report findings, and emit a pass/fail verdict WITHOUT editing any code. Use during the review-gate stage of the dev→review→test loop.

Strengthen a module's test suite with the pi coding agent — add missing edge-case, error-path and regression tests, then run them to confirm they pass. Use for a standalone test-hardening step (script or plugin) outside the main task gate.

Оформить задачу развития экосистемы как tasks.md-спеку для spec-runner: из кластера weekly self-review Robin, находки fleet-отчёта или явного запроса пользователя — в managed-спеку (draft) PR-ом в репо-владелец изменений. Запускать: «оформи спеку», «преврати кластер/находку в задачи», «заведи задачи по этому провалу», «spec bridge».

Периодическая проверка флота репозиториев ATP-экосистемы: незакоммиченные файлы, ahead/behind относительно upstream, открытые PRs, issues, security alerts по каждому репо. Формирует markdown-отчёт для prograph-vault derived/fleet/ и доставляет его PR-ом. Запускать: по просьбе «проверь флот / состояние репозиториев / fleet check», при старте рабочей сессии и по расписанию (например, ежедневно утром).

Libretto VM skill pack. Activate on any `libretto` command, legacy `prose` command, .libretto files, or Libretto mentions; orchestrates multi-agent workflows.

Провести discovery-интервью со стейкхолдером (фрейм customer или engineer) и выдать структурированный discovery-brief по контракту v0.1 с обязательным coverage-gate. Use when the user says "discovery interview", "проведи discovery", "интервью заказчика/инженера", wants to elicit product goals, functions, or system constraints from a stakeholder, or needs a discovery brief feeding Gate BR/FRD/0b/0a.