claudekit

Use when reviewing the architecture dimension of a written plan. Dispatched primarily by plan-review-architecture (via plan-review). Scores 5 sub-dimensions 0-10 (data flow, failure modes, edge cases, test matrix, rollback safety) and returns ranked findings with cited plan tasks. <example> Context: A plan has been written and is about to be implemented. user: "Run plan-review on the cache-invalidation plan." assistant: "Dispatching the architect agent to score the architecture dimension while the experience-reviewer runs in parallel." </example> <example> Context: A migration plan needs an architecture-only pass. user: "I just need an arch review on this — skip the UX review." assistant: "Dispatching the architect agent directly." </example>

Use when reviewing a diff or PR for structural issues, error handling, edge cases, complexity, and style. Dispatched primarily by code-review-loop. Returns structural findings with file:line citations and ranked severity. Pairs with security-auditor for sensitive paths. <example> Context: A PR is ready for first-pass review. user: "Review my charge-endpoint PR before I tag humans." assistant: "Dispatching the code-reviewer agent to find structural issues, error-handling gaps, and complexity hotspots." </example> <example> Context: A refactor PR needs a sanity check. user: "Sanity-check this refactor PR." assistant: "Dispatching the code-reviewer to confirm behavior preservation and look for unintended changes." </example>

Use when reviewing the experience dimension of a written plan (UX + DX). Dispatched primarily by plan-review-experience (via plan-review). Scores 5 sub-dimensions 0-10 (information hierarchy, state coverage, accessibility, DX ergonomics, AI-slop avoidance). <example> Context: A plan with both UI and API changes needs review. user: "Run plan-review on the dashboard plan." assistant: "Dispatching the experience-reviewer agent in parallel with the architect to cover UX and DX in one pass." </example> <example> Context: A new public API surface is being added. user: "Review the DX of the new webhook API plan." assistant: "Dispatching the experience-reviewer to score DX ergonomics, error copy, and discoverability." </example>

Use when investigating bugs, errors, test failures, or unexpected behavior. Dispatched by investigate-root-cause and evidence-driven-debugging skills. Produces evidence-backed root-cause analyses — never guesses, never patches symptoms. <example> Context: An API endpoint is returning intermittent 500s. user: "The /api/users endpoint is throwing 500s sometimes." assistant: "Dispatching the investigator agent to gather evidence, write a hypothesis, and prove or refute it before any fix." </example> <example> Context: Tests passed locally but fail in CI. user: "My tests pass locally but CI is red." assistant: "Dispatching the investigator to find the env diff between local and CI and produce a hypothesis." </example>

Use when decomposing a spec into an executable plan. Dispatched primarily by the write-plan skill. Produces a numbered task list with file paths, exact test commands, dependency annotations, acceptance criteria per task, and a Risks section. <example> Context: An approved spec exists; implementation hasn't started. user: "Turn the auth-rotation spec into a plan we can execute." assistant: "Dispatching the planner agent to produce a numbered task list with file paths, test commands, and rollback notes." </example> <example> Context: A previous plan was rejected during plan-review for being too vague. user: "Re-plan the migration; the reviewers said it had no acceptance criteria." assistant: "Dispatching the planner agent to rebuild the plan with falsifiable acceptance lines per task." </example>

Use when investigating dependency bloat, security advisories, supply-chain risk, upgrade planning, or before adding a new third-party package. Activate for keywords like "deps", "dependencies", "package.json", "requirements.txt", "Cargo.toml", "audit", "CVE", "stale package", "do we use", "what depends on", "transitive dep". Produces a written audit with import-graph evidence — never trust scanner output without verifying call sites.

Use when opening a PR for review or when receiving review feedback. Activate for keywords like "code review", "PR review", "request review", "review feedback", "address comments", "reviewer said". Covers both ends of the loop: preparing a reviewable PR and acting on feedback rigorously. Always engage with every comment -- never dismiss feedback by silently ignoring it.

Use during active debugging when you have a hypothesis to test or need to instrument a running system. Activate for keywords like "debug", "instrument", "log", "trace", "breakpoint", "what's happening at runtime", "production behavior". Pair to investigate-root-cause for the doing-it phase. Always record what you observed -- never debug entirely "in your head" without leaving evidence behind.

Use when implementing a non-trivial feature, migration, or refactor that would otherwise be a single large change. Activate for keywords like "feature flag", "incremental", "vertical slice", "migration", "rollout", "behind a flag", "ship small". Enforces vertical slicing, feature-flagged rollout, and refactor-with- evidence (behavior-preserving changes proved by test/perf deltas). Always ship the smallest reversible change -- never bundle unrelated improvements.

Interactive setup wizard for claudekit. Scaffolds rules, hooks, and MCP server configs into the user's project. Run /claudekit:init to configure. Use when setting up a new project with claudekit or reconfiguring an existing one.

Creative exploration mode — divergent thinking, multiple alternatives, structured trade-offs before any code

Critical analysis mode — find issues first, severity-tagged findings, actionable suggestions

Thorough investigation mode — completeness over speed, evidence-cited, confidence levels named

Code-focused execution mode — minimal prose, action-oriented updates, follow established patterns

Compressed output mode — minimal prose, code-first, no preambles