yuri-semenenko/ai-engineering-workspace
Agent Claude Code
Independent second read of a change, for the review step of explore -> implement -> review. Dispatch it when the code was written in this conversation and you want findings from something that has not seen the reasoning behind it. Returns candidate findings with file:line and a severity guess, never a verdict and…
yuri-semenenko/ai-engineering-workspace
Agent Claude Code
TEMPLATE for a project-specific implementation agent. Copy this file, rename it, and fill in the placeholders to encode ONE project's stack, conventions, and workflow into a reusable agent. Example trigger description below — rewrite it for your project.\n\nExamples:\n\n- User: "Create a new page that lists with…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Scaffold an Architecture Decision Record (ADR) for a concrete decision that's already been made or is being made now. Use when the user says "ADR for X", "write up that decision", "ADR по ", or wants to record a choice for future readers. Different from /rfc — RFC explores options; ADR commits to one.
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Orient fast in unfamiliar code — a new repo, a subsystem you don't own, or a PR touching code you don't know — and produce a short, high-signal map: what it is, entry points, architecture sketch, domain glossary, key seams, risky areas, how to run and test, known tradeoffs. Use for "map this repo", "orient me in this…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Plan and create small, logically-grouped git commits per the persona's git conventions — propose a commit plan first, wait for approval, then commit in atomic units. Use when the user asks to "commit", "закоммить", "make commits", "split this into commits", "сделай коммиты", or "commit plan". Enforces the…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Scan an entire codebase (or a chosen subtree) for over-engineering — premature abstraction, speculative generality, needless layering, wrapper-only modules, dead config/flags, indirection without payoff — and return a prioritized, deletion-oriented report. Use for "audit complexity", "find over-engineering", "where…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Collect the deliberate-simplicity tradeoff annotations left in the code (// TRADEOFF(ceiling: ...; upgrade: ...): ...) across the repo into a single ledger — location, ceiling, upgrade path, note. Use for "debt ledger", "list tradeoffs", "what shortcuts did we take", "debt-ledger". Pairs with the TRADEOFF convention…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Systematic root-cause debugging methodology — reproduce, isolate, hypothesize, verify, fix the cause not the symptom. Use when something is broken and the cause is unknown: "debug this", "why is X failing/broken", "find the root cause", "разберись почему падает", "найди причину бага", "debug по ". Not for confirming a…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Rewrite text to remove the telltale patterns of AI-generated writing. Use only when the user explicitly asks to humanize / de-AI a piece of text ("humanize this", "убери признаки AI", "make it sound less AI / more natural", "this sounds robotic") — not on every text edit. Covers five pattern families: inflation (fake…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Apply the "laziest solution that actually works" enforcement ladder when writing or proposing code — question whether the work needs to exist (YAGNI), reach for stdlib / native platform / already-installed deps before custom code, prefer one line over fifty. Supports intensity lite | full (default) | ultra. Use when…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Turn a risky migration or refactor into a safe, incremental, reversible plan that keeps user-visible behavior stable — framework upgrades, design-system swaps, API version migrations, legacy-to-modern replacement. Produces the invariant, the current-to-target seam, a migration shape (strangler / adapter /…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Shape a module so a small, stable interface hides substantial implementation — deep modules, information hiding, and the test for whether an abstraction earns its keep. Use when designing or reviewing a module or interface boundary, extracting a helper or wrapper, or deciding whether an abstraction is worth it…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Review a pull request and classify every finding as Critical / Important / Optional per the user's persona framework. Use when the user asks for a "PR review", "review this PR", "проревьюй PR", or wants PR-level triage. Different from /review (generic GitHub PR review) and from /code-review (working-diff bug scan) …
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Generate a pull-request description (the "PR comment") that fills the repo's own PR template, then output it as a copy-pasteable raw markdown code block. Use when the user asks to "prepare a PR comment / description", draft the body for gh pr create, or "подготовь PR-коммент".
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Second-pass re-review of a PR that already has review comments. Re-reads the diff and new commits against the existing open threads, marks each as addressed / partial / not addressed, resolves the addressed ones, and then either approves (only when everything is clean) or drafts new inline comments. Use when the user…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Gather evidence from primary sources to decide a concrete technical question — compare libraries, frameworks, APIs, or approaches — and produce a cited decision matrix with a recommendation and its uncertainty, ready to hand to an RFC or ADR. Use for "research X", "compare A vs B", "which library should we use"…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Draft an architecture RFC using the user's canonical 10-section format (Problem → Context → Constraints → Assumptions → Options → Trade-offs → Recommendation → Risks → Migration Strategy → Open Questions). Use when the user asks for an "RFC", "design doc", "architectural proposal", or "RFC по ". Optimized for the…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Run a security hardening pass using the persona's validated workflow — recon, then a numbered remediation checklist for approval, then atomic commits with a typecheck gate after each task. Use when the user asks for a "security pass", "harden this", "security hardening", "пройди по безопасности", or a staged…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Write a minimal spec — goal, non-goals, acceptance criteria, open questions — before implementing a feature or change. Use when the user asks to "spec this", "напиши спеку", "what does done mean here", "acceptance criteria for X", or before starting a non-trivial implementation without a clear definition of done.…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Onboarding entrypoint for this kit. Explains what the persona wizard does and hands you the exact command to generate your own persona (persona.md, CLAUDE.md, recommended-skills.md) from the shared templates. Use when the user says "/start", "get started", "onboard me", "set up my persona", or is new to the workspace.…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Testing checklist — test pyramid, what to test vs skip, DAMP over DRY, the TDD red-green-refactor loop for new code, failing-test-first for bug fixes, Vitest / React Testing Library patterns. Use when writing or reviewing tests, deciding test coverage for a change, driving new code test-first, fixing a bug (regression…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Web performance checklist — Core Web Vitals (LCP/INP/CLS), TTFB diagnosis, and frontend levers (JS/CSS/fonts/images/network/rendering) plus backend (DB N+1, API latency, caching, infra). Use when optimizing performance, diagnosing a slow page or unresponsive interaction, setting a performance budget, or reviewing a…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Web application security checklist — OWASP Top 10 + LLM Top 10, threat modeling (STRIDE), auth/authz, input validation, secrets, security headers, CORS, data protection, and supply-chain hygiene. Use when reviewing security, hardening a feature, threat-modeling, or before merging security-sensitive changes (auth, user…
yuri-semenenko/ai-engineering-workspace
Skill Claude CodeCodex
Use for architectural decisions, RFCs, ADRs, system design, trade-off analysis, migration planning, and complex technical decisions that need Staff-level reasoning. Returns a structured recommendation rather than implementation code.