alexei-led

Browser automation for rendered UI exploration, validation, screenshots, recordings, and end-to-end flows. Use when a task needs an actual browser or rendered DOM: inspect UI state, click/fill forms, debug frontend behavior, capture evidence, verify a…

Create normal git commits with logical grouping. Use when committing, saving changes, creating commits, or grouping work into commits. For drafting or editing commit messages without committing, use writing-style. NOT for amending, rebasing, force-pushing, or…

Write, rewrite, or update project docs from implementation facts - README front pages, user guides, configuration references, architecture docs, evaluations, agent instructions, and code comments. Use when writing or improving project docs for clarity,…

Diagnose cc-thingz plugin installations with the bundled read-only doctor. Use for missing installed skills, helpers, or hook files, stale package versions, duplicate packages, or migration from old cc-thingz packages. NOT for broad agent configuration audits…

Prepare, publish, or repair a software release and write its release notes. Use for cutting a versioned release or editing release notes. NOT for updating an installed package, publishing articles, or discussing a release; for general documentation use…

Use when asked to lint, audit, review, or score AI-facing instruction files such as SKILL.md, AGENT.md, AGENTS.md, CLAUDE.md, platform body.md files, prompt files, rules, policies, and agent-facing references. NOT for plugin manifests, application code…

Create, split, slim, or rewrite repository skills. Use when adding a new `src/skills/<name>/` skill, editing a skill description, frontmatter, references, overlays, or plugin placement, or tightening routing between neighboring skills. NOT for score-only…

Write or edit commit messages, PR/MR descriptions, issues, review comments, and technical discussions. Use for concise, clear engineering communication grounded in facts, without filler or boilerplate. NOT for creating commits (committing-code), project docs…

Turn approved architecture-design artifacts, with supporting review findings when present, into an incremental implementation/refactoring plan. Use when asked to plan a refactor, sequence remediation, or implement an approved target architecture. Produces a…

Run a repeatable, evidence-based architecture review of an existing codebase. Use when asked to assess modularity, coupling, cohesion, dependency direction, circular dependencies, blast radius, fragile seams, shallow modules, testability, ownership…

Gather deterministic architecture-fitness evidence with archfit: full checks, scorecards, deltas, SARIF/JSON findings, tool coverage, and agent_tasks. Use when a repo has .archfit.yaml or the user wants architect to combine its review with archfit. Produces…

Gather TypeScript/JavaScript architecture evidence with dependency-cruiser, madge, knip, tsc, ESLint, and the package manager. Use when a TS/JS repo needs module graphs, workspace/package boundaries, import cycles, dependency direction, dead code, type…

Design a modular architecture from requirements, intended architecture notes, or an approved redesign brief. Use when asked to create architecture docs, define module boundaries, integration contracts, test specifications, or a target architecture before…

Assign architecture scores against the fixed rubric. Use when scoring or re-scoring architecture dimensions, choosing a band, setting confidence, or checking that a score is defensible. Enforces band-matches-value, evidence-per-score, and low-confidence caps.…

Turn approved architecture-design artifacts, with supporting review findings when present, into an incremental implementation/refactoring plan. Use when asked to plan a refactor, sequence remediation, or implement an approved target architecture. Produces a…

Run a repeatable, evidence-based architecture review of an existing codebase. Use when asked to assess modularity, coupling, cohesion, dependency direction, circular dependencies, blast radius, fragile seams, shallow modules, testability, ownership…

Manual pin: run this turn on sonnet at xhigh effort. Use as /router:medium <prompt>.

Point Claude Code at the router gateway. Adds `model` with the 1M window, `ANTHROPIC_BASE_URL`, tool search and the `/model` picker row to the user settings, and offers the status line.

Show which model served the last turn, whether Jev routing is active, and the routing config.

Manual pin: run this turn on opus at xhigh effort. Use as /router:high <prompt>.

Manual pin: run this turn on sonnet. Use as /router:low <prompt>.

Manual pin: run this turn on haiku. Use as /router:micro <prompt>.

Set up a small, verifiable implementation plan.

Build and verify coding-agent skill bundles with **Agent Bundler**. Use when configuring agentbundle.json, creating or updating skills-repository, bundle, or Claude-plugin sources, selecting targets or packages, applying target overlays, generating output,…

Keep tool calls safe

Prepare a deterministic package release.

Review a change against the repository rules.

Use archfit for architecture fitness work. Use when installing archfit, creating or reviewing .archfit.yaml, configuring Go, Python, TypeScript, or Rust checks, adding CI usage, interpreting findings, fixing architecture drift, or deciding on baselines and…

Use to evaluate archfit itself against the book-alignment prompt, the test corpus, CLI UX, docs, and skill quality. Use when auditing archfit coverage vs Balancing Coupling, running corpus sweeps, comparing deterministic archfit output with semantic…

Full release lifecycle — version bump, CHANGELOG, rich release notes, tag, publish. Use when user says "release", "tag and release", "publish version", "cut a release", "new version".

Query Spot/preemptible VM prices, savings and interruption risk across AWS, GCP and Azure with the spotinfo CLI. Use this whenever the user asks which instance type is cheapest, what a Spot instance costs, what the interruption risk is, whether arm64 beats…

Use before non-trivial coding work to pick a task-appropriate working mode — debug, fix, review, test-first, refactor, optimize, migrate, upgrade, integrate, spike, or unstuck — then follow its checklist. Modes are procedures, not personas. Not needed for…

Run a pi-fusion review: several models answer in parallel, then one synthesis step returns a single report. Use when the user says "invoke fusion", "run a fusion panel", "discuss this through fusion", "get multi-model opinions", "panel review", "compare…

Plan, run, inspect, pause, resume, adopt, diagnose, reconcile, retire, or recover a checked Markdown implementation plan through pi-plan-exec. Use when a plan-exec run is active, stuck, failed, detached after reload, cancel-pending, abandoned after a Pi…