ivanestar

architecture-doc-writer

How to write and maintain EITR's architecture documentation (docs/architecture/) - arc42-lite structure for the living system description, Nygard-format ADRs for individual decisions. Load before editing anything under docs/architecture/, or when asked to document a new architectural decision.

ast-template-engineer

Protocol for safely modifying TypeScript AST templates, enforcing compilation checks and a strict two-strike fallback rule. Trigger on requests like "edit the template", "modify packages/engine/src/plan/templates", "update the AST generator".

axis-closure-matrix

Produces a research-backed Core/Extended gap matrix for exactly ONE named EITR architecture axis (e.g. "CPOM base for Python", "CI/CD generator", "AI-agent layer") at a time, ending in a checkable Definition of Done and explicit re-open triggers instead of an open-ended audit. Trigger only when the user explicitly names a specific axis and asks to research/score/close it — phrases like "матрица по оси X", "closure-план для X", "проверь ось X", "что осталось по X". Never self-invoke to propose a new axis on your own initiative.

Code & Development Listed

commit-writer

Rules and guidelines for structuring and formatting Git commits in the EITR repository, covering Conventional Commits scoping, deterministic body triggers, and the mandatory 23:00 OpSec timestamp policy. Trigger on requests like "write a commit message", "commit this", "как закоммитить", "напиши commit message".

framework-quality-audit

On-demand, unbiased multi-angle quality audit of an EITR-generated test framework for any language + automation-tool combination (e.g. TypeScript+Playwright, Python+pytest, Java+Selenium, C#+Playwright). Scores a fixed set of 19 categories on a 1-10 scale with objective file-referenced evidence, lists ALL found gaps (not just top picks) and concrete killer-feature ideas, and publishes a deterministically-structured HTML Artifact report so every audit looks and reads the same regardless of stack or date. Trigger on requests like "аудит фреймворка", "оцени генерацию", "quality audit for <language>+<tool>", "проведи ревью качества".

Testing & QA Listed

protocol-123

Strict 9-phase multi-agent autonomous engineering and bugfixing pipeline triggered by '123' or 'по 123'. Enforces SDD-first specification, multi-agent plan review, user sign-off gateway, TDD-driven execution (Red-Green-Refactor), multi-agent code/test review, two-strike self-healing, and QA verification.

protocol-456

6-phase lightweight engineering pipeline (Phases 0-5) triggered by '456', 'по 456', 'таска по 456', or 'исправь по 456' for fast-track tasks (everyday bugfixes, small features, or pre-scoped tracks). Slashes token and time overhead by 80-85% compared to Protocol 123 via compact SDD planning, on-demand web research, in-session TDD, single-reviewer risk audit, and deterministic runner scripts/protocol-456.mjs. Pre-scoped tracks skip Phases 1 & 2 directly to Phase 3. Never self-invoke.

sdd-plan-writer

Writes a self-contained, step-by-step Spec-Driven Development (SDD) implementation plan for an AI coding agent to execute against an already-confirmed batch of work items (bugs, gaps, features) — exact file paths per task, EARS-style trigger/response acceptance criteria, a runnable verification command per task, explicit [NEEDS CLARIFICATION] markers instead of guessing, and dependency-ordered phases. Trigger on requests like "напиши SDD-план", "распиши план по шагам для агента", "сделай implementation plan для батча X" once the work items are already identified. Do NOT trigger for open-ended "what should we build" ideation — that needs investigation first, not this skill.

stack-scaler

Step-by-step guide for extending the Eitr codebase to support a new programming language and/or E2E test automation tool. Covers the full pipeline from CLI questionnaire through plan.ts generator registry to template generation and tests.

strategic-architecture-advisor

Acts as a strategic architect-advisor for EITR's own product direction (not a generated project's) - reviews current architecture against fresh industry research, hunts for non-obvious synergies between backlog ideas, designs the SDET onboarding flow across greenfield/brownfield x docs/no-docs x tests/no-tests, and records business-language questions only the user can answer. Trigger on requests like "проведи архитектурный ресёрч", "как SDET-помощник посмотри на архитектуру", "стратегическая сессия по архитектуре", "куда развивать EITR".