ai-kit

Explain the current git branch's changes in plain product language for a non-technical audience — no file paths, no code references, no jargon. Reads the branch diff, commits, and available docs, then produces a short human briefing about what changed and why it matters. Use when the user invokes /explain-branch-changes, asks what a branch does in simple terms, needs a product summary of changes for a PM / designer / stakeholder / support, or says things like 'объясни ветку простыми словами', 'что поменялось для пользователей', 'explain this branch to a non-developer'. Works with any codebase, language, and project structure. NOT for code review or technical diff analysis — use load-branch-changes for that.

Sync README with current agents/skills, validate plugin conventions, sync the Codex plugin directory (plugins/ai-kit), bump version, and commit. Use when the user says "/review-and-release" or wants to prepare a release after adding, removing, or changing agents or skills.

Perform 12-Factor App compliance analysis on any codebase. Use when evaluating application architecture, auditing SaaS applications, or reviewing cloud-native applications against the original 12-Factor methodology.

Automated technical website audit via the squirrelscan CLI: crawls a live or local site and checks 230+ rules across SEO, performance, security, technical, content, and 16 other categories, returning an LLM-optimized report with a health score, broken links, meta-tag analysis, and actionable fixes (with an optional confirm-first fix loop). Use for a fast, tool-driven scan of a website or webapp. For manual SEO strategy and content review without a crawler, use the seo-audit skill instead.

Check browser support for web platform features (CSS, JS APIs, HTML) against the project's actual browser targets. Resolves the project's browserslist config, queries caniuse/MDN compatibility data, reports Baseline status, and recommends fallbacks or progressive enhancement. Use when asking "can I use X", choosing between modern and legacy approaches, reviewing code for unsupported features, or auditing browser compatibility.

Analyzes code diffs and files to identify bugs, security vulnerabilities (SQL injection, XSS, insecure deserialization), code smells, N+1 queries, naming issues, and architectural concerns, then produces a structured review report with prioritized, actionable feedback. Use when reviewing pull requests, conducting code quality audits, identifying refactoring opportunities, or checking for security issues.

Скилл для очеловечивания русскоязычного текста. Убирает признаки AI-генерации, делает текст живым. Используй ВСЕГДА, когда пользователь просит: очеловечить текст, убрать следы нейросети, сделать текст живым/естественным, переписать как человек, humanize на русском, убрать канцелярит, убрать водянистость, сделать текст менее формальным. Также используй если пользователь вставляет русскоязычный текст и говорит что-то вроде 'перепиши', 'сделай лучше', 'звучит как робот', 'слишком искусственно'. Работает ТОЛЬКО с русским языком. Для английского используй оригинальный humanizer. НЕ используй для: перевод, написание с нуля, грамматика, код.

Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, passive voice, negative parallelisms, and filler phrases.

Write, fix, or review JSDoc documentation for JavaScript and TypeScript code. Detects whether the project uses typed JSDoc (plain JS with checkJs), TypeScript (docs without type duplication), or a doc generator (TypeDoc, JSDoc CLI), and documents accordingly. Use when adding JSDoc comments, documenting a public API, fixing doc/signature drift, or setting up doc generation.

Load current git branch changes (diff, commits, changed files) into session context. Use when the user invokes /load-branch-changes to review, understand, or focus on the current branch's changes compared to the base branch.

Audit a module, package, or subsystem for compatibility with its consumers after a change. Compares baseline vs candidate public surface, classifies every change (additive, source-breaking, runtime-breaking, behavioral, data-breaking, dependency-breaking), verifies affected consumers, and emits a compatibility matrix with a compatible / conditionally-compatible / breaking verdict plus a migration path. Use when refactoring a shared module, upgrading a library's API, extracting a package, or answering "will this change break anyone".

Use when building Next.js applications with App Router, server components, or server actions. Invoke to configure route handlers, implement middleware, set up API routes, add streaming SSR, write generateMetadata for SEO, scaffold loading.tsx/error.tsx boundaries, or deploy. Triggers on: Next.js, App Router, RSC, use server, Server Components, Server Actions, React Server Components, generateMetadata, loading.tsx, Next.js deployment, Vercel, Next.js performance.

Reset accumulated permissions in .claude/settings.local.json. Use when the user says "/reset-permissions" or asks to clear, reset, or clean up Claude Code permissions. Supports full reset (default) or selective reset keeping read-only permissions (--keep-read).

Run a comprehensive Go code review on git-changed files using the golang-pro agent with Context7 library docs

Manual, strategic SEO review and methodology — no tooling required. Use when the user wants to audit, diagnose, or improve SEO and ranking, or mentions "SEO audit," "technical SEO," "why am I not ranking," "on-page SEO," "meta tags review," "E-E-A-T," or "SEO health check." Covers crawlability, on-page, content quality, and keyword strategy, ending in a prioritized report. For an automated, crawler-driven scan with a health score, use the audit-website skill instead.

Keep code comments signal-only in any programming language — never write comments that restate the code, and delete existing noise comments instead of rewriting them. Use when writing or generating code (to avoid producing filler comments), when asked to clean up, simplify, or remove comments, or when reviewing code for comment noise. Preserves doc comments (JSDoc, docstrings, javadoc, rustdoc, godoc), "why" comments, ticket/ADR anchors, and deprecation notices.

Prove that a unit, integration, or end-to-end test genuinely guards its stated behavior by running a few targeted fault probes (reachability, sensitivity, oracle validity, reliability) around the code under test — instead of trusting coverage. Use during active test development or test review: after writing or changing a test, when verifying a test actually catches real bugs, when a test "passes but might not test anything", when a suite feels untrustworthy, or to health-check the tests affected by the current branch. Supports an isolated git-worktree mode scoped to affected cases so production code is never touched.

Audit and update Go module dependencies using native go toolchain commands (go list, go get, go mod tidy) plus govulncheck for security. Auto-applies patch updates silently, asks before minor, and confirms each major individually. Cross-checks release notes via gh/Context7 and vulnerabilities before each batch. Handles v2+ module path changes (gopkg.in/, github.com/.../v2). Use when the user asks to update dependencies, bump modules, check outdated modules, or audit security in a Go project.

Audit and update Node.js/JavaScript project dependencies using native package-manager commands (npm, pnpm, yarn, bun). Auto-applies patch updates silently, asks before minor, and confirms each major individually. Cross-checks changelogs via Context7 and security advisories before each batch. Use when the user asks to update dependencies, bump packages, check outdated packages, or audit security in a JS/TS project.