ai-dev-tools

Plugin ai-dev-tools from marketplace ai-dev-toolkit in uakdemir/ai-dev-toolkit. Commands read from the repo's manifest — 19 days ago.

Skills (16)

• api-contract-guard

  Use when the user wants to define module API boundaries, enforce that consumers import through barrel files, generate contract tests for module encapsulation, create index/barrel files for modules, or prevent AI agents from importing internal module paths — even if they don't use the exact skill nam

• brainstorm

  Use when the user wants to brainstorm a feature, a change or an idea into a written spec — runs superpowers:brainstorming so that it ends with a committed spec under docs/superpowers/specs/, and prints the /review-doc command for it. The first step of the brainstorm → review-doc → implement → review

• changelog-from-commits

  Use when the user wants to generate release notes, create a changelog, produce a changelog from git history, document what changed between versions, summarize commits for a release, or group changes by type — even if they don't use the exact skill name.

• consolidate

  Unify AI configs and linting rules across a monorepo. Diffs AI configs across projects, wires lint config inheritance, and offers direct unification where inheritance isn't available. Use /consolidate ai for AI configs, /consolidate lint for linting, /consolidate all for both.

• convention-enforcer

  Use when the user wants to enforce coding conventions, prevent AI agent drift, harden linter rules, generate convention-based structural tests, detect naming/error handling/import/logging inconsistencies, or analyze a codebase for convention violations — even if they don't use the exact skill name.

• create-artifact

  Use when the user asks to make, change, publish or finalize an engineering artifact: one HTML page that shows a decision, a system or a screen to a person. Applies to a decision page, a diagram page, a mockup page, an HTML artifact, and the conversion of a Claude Design board or canvas (`.dc.html`)

• document-for-ai

  Use when the user wants to create, migrate, audit, or maintain AI-optimized documentation, restructure existing docs for AI consumption, generate CLAUDE.md files, create doc indexes, or improve AI agent efficiency through better documentation — even if they don't use the exact skill name.

• help

  Show available commands and usage for the ai-dev-tools plugin

• implement

  Use when the user wants to execute a written implementation plan or implement directly from a spec — decides whether a spec needs a plan, generates a task graph, splits the plan among at most three implementers, and dispatches with quality overrides. Invoke with /implement <path>.

• refactor-to-layers

  Use when the user wants to enforce architectural layers within a module or project, add dependency direction rules, reduce AI agent context through structural boundaries, introduce provider/dependency injection patterns, or generate structural tests that enforce layer compliance — even if they don't

• refactor-to-monorepo

  Use when the user wants to split a monolith into modules, identify module boundaries, analyze code coupling, plan a monorepo migration, evaluate monorepo tooling, or reduce codebase size for better AI agent efficiency — even if they don't explicitly say monorepo.

• review-code

  Use when reviewing recent commits for bugs, architecture violations, spec drift, security issues, and verification gaps. Supports single-pass review and iterative review-fix-verify cycles. Invoke with /review-code <commit-count|git-ref>.

• review-doc

  Use when reviewing analysis specs, design documents, or implementation plans for completeness, accuracy, and implementability. Supports single-pass review (--max-iterations 1) and iterative review-fix cycles. Invoke with /review-doc <path1> [path2 ...] or /review-doc <directory/>.

• scaffold

  Use when the user wants to bootstrap a project from scratch, scaffold a new project, generate the initial CLAUDE.md/.claude/ layout for a fresh repo, or add a new package/feature to an existing scaffolded project — even if they don't use the exact skill name. Supports --bootstrap (fresh project) and

• session-handoff

  Use when the user wants to generate a handoff document for the next AI session, capture session progress, save context between sessions, create a session summary, preserve decisions and pending work, wrap up a session, or indicate they are done for now — even if they don't use the exact skill name.

• test-audit

  Use when the user wants to audit test quality, find coverage gaps, detect flaky tests, assess assertion quality, identify missing edge cases, prioritize test improvements, or evaluate test suite health — even if they don't use the exact skill name.

Subagents (3)

• high-effort

  Role-neutral agent pinned at `high` reasoning effort, whatever effort the dispatching session is at. Use it only when a skill or a prompt names it: /review-code, /review-doc and /implement dispatch it when the run's tier or a flag selects `high`. It pins no model, so pass `model` on the Agent call.

• max-effort

  Role-neutral agent pinned at `max` reasoning effort, whatever effort the dispatching session is at. Use it only when a skill or a prompt names it: /review-code, /review-doc and /implement dispatch it when the run's tier or a flag selects `max`. It pins no model, so pass `model` on the Agent call.

• xhigh-effort

  Role-neutral agent pinned at `xhigh` reasoning effort, whatever effort the dispatching session is at. Use it only when a skill or a prompt names it: /review-code, /review-doc and /implement dispatch it when the run's tier or a flag selects `xhigh`. It pins no model, so pass `model` on the Agent call

Similar plugins