skills.json

{
  "version": 1,
  "skills": [
    {
      "name": "agent-native-architecture",
      "description": "Build applications where agents are first-class citizens. Use this skill when designing autonomous agents, creating MCP tools, implementing self-modifying systems, or building apps where features are outcomes achieved by agents operating in a loop.",
      "path": "skills/agent-native-architecture",
      "version": "2.0.0"
    },
    {
      "name": "api-design",
      "description": "Design and review robust API and interface contracts. Use when creating or changing HTTP endpoints, GraphQL/RPC-style APIs, webhooks, SSE/event streams, exported DTOs/types, service-layer boundaries, CLI JSON/stdout/exit-code contracts, versioning/deprecation plans, or any compatibility-sensitive boundary between consumers and providers.",
      "path": "skills/api-design",
      "version": "1.0.0"
    },
    {
      "name": "audit-skill",
      "description": "Security audit for agent skills \u2014 prompt-injection and exfiltration scanning with an A\u2013F trust score. Use when reviewing a skill for security, auditing a skill before installation, checking for prompt injection, or when the user says 'audit skill', 'check skill security', 'trust score', 'is this skill safe'. On-demand via /audit-skill.",
      "path": "skills/audit-skill",
      "version": "1.0.4"
    },
    {
      "name": "blind-spots",
      "description": "Find the gaps in a user's understanding of a code change \u2014 before it gets built, or after an agent has built it. Two modes \u2014 scope maps blast radius and unknown unknowns; quiz questions the user on a diff an agent just made. Use when the user asks for a blind spot pass, wants the scope of a proposed change, or asks to be quizzed on a change. Triggers on \"blind spot pass\", \"find my blind spots\", \"unknown unknowns\", \"help me understand this change\", \"quiz me on this change\".",
      "path": "skills/blind-spots",
      "version": "1.0.4",
      "triggers": [
        "blind spot pass",
        "find my blind spots",
        "unknown unknowns",
        "help me understand this change",
        "quiz me on this change"
      ]
    },
    {
      "name": "brainstorming",
      "description": "Use this when requirements are ambiguous, multiple approaches are plausible, or trade-offs need discussion before planning or implementation. Clarifies WHAT to build through one-question-at-a-time collaboration. Can be skipped when requirements are already explicit and well constrained.",
      "path": "skills/brainstorming",
      "version": "2.0.0"
    },
    {
      "name": "caveman",
      "description": "Ultra-compressed communication mode, persistent across turns once active. Use when the user says \"caveman mode\", \"talk like caveman\", \"use caveman\", \"less tokens\", asks to compress agent responses for the rest of the session, or invokes /caveman. Stop with \"stop caveman\" or \"normal mode\".",
      "path": "skills/caveman",
      "version": "1.0.1"
    },
    {
      "name": "compound-docs",
      "description": "Capture solved problems as categorized documentation with YAML frontmatter for fast lookup. Use when a solution is confirmed and should be preserved for future retrieval.",
      "path": "skills/compound-docs",
      "version": "1.0.1",
      "allowed-tools": [
        "Read",
        "Write",
        "Bash",
        "Grep"
      ]
    },
    {
      "name": "create-cli",
      "description": "Design command-line interface parameters and UX: arguments, flags, subcommands, help text, output formats, error messages, exit codes, prompts, config/env precedence, and safe/dry-run behavior. Use when you\u2019re designing a CLI spec (before implementation) or refactoring an existing CLI\u2019s surface area for consistency, composability, and discoverability.\n",
      "path": "skills/create-cli",
      "version": "1.0.0"
    },
    {
      "name": "deep-research",
      "description": "Use when a task needs direct web-backed research with citation-ready synthesis \u2014 the user wants the answer, not a commissioned research program. For commissioning multi-model or externally-executed research programs, or verifying reports produced elsewhere, use research-architect instead \u2014 this skill is its execution backend.",
      "path": "skills/deep-research",
      "version": "2.3.3"
    },
    {
      "name": "design-critique",
      "description": "Audit implemented UI against a 37-pattern slop catalog and return ranked, scoped findings with named alternatives. Use when the user asks to review their UI, audit a design, check for AI-generated tells, critique the visuals, or asks 'does this look AI-generated'. For rule-compliance, accessibility, or UX audits use web-design-guidelines instead.",
      "path": "skills/design-critique",
      "version": "1.0.1"
    },
    {
      "name": "design-md",
      "description": "Read, write, lint, diff, and export DESIGN.md files using the Google @google/design.md format. Use when the user mentions DESIGN.md, design tokens, extracting a design system, linting design tokens, exporting tokens to Tailwind, DTCG, or CSS variables, or when authoring a fresh design-system reference for a project.",
      "path": "skills/design-md",
      "version": "1.0.2"
    },
    {
      "name": "diagnose",
      "description": "Disciplined debugging loop for hard bugs and performance regressions. Use when the user says \"diagnose this\", \"debug this\", \"this is broken/throwing/failing\", reports a bug whose cause is non-obvious, or describes a performance regression. For completion-time evidence checks use verify-before-complete; for new-test methodology use test-strategy; for post-hoc code review use local-review.",
      "path": "skills/diagnose",
      "version": "1.0.4"
    },
    {
      "name": "error-handling-review",
      "description": "Review changed code for silent failures and inadequate error handling \u2014 empty or over-broad catch blocks, errors logged-and-swallowed, unjustified fallbacks, null/optional-chaining that hides failures, and retries that exhaust silently. Use when the user asks to review error handling, check catch/try-except blocks, audit fallback behavior, or hunt for swallowed errors. Not general diff review (local-review) or security scanning (secure-code).",
      "path": "skills/error-handling-review",
      "version": "1.0.0"
    },
    {
      "name": "fetchmd",
      "description": "Fetch webpages or local HTML and convert to clean, token-efficient markdown. Use when you need to ingest web content for summarization, RAG, research, or any AI workflow that requires readable text from URLs or HTML files.\n",
      "path": "skills/fetchmd",
      "version": "1.0.0"
    },
    {
      "name": "find-skills",
      "description": "Discover and install agent skills. Use when the user asks \"how do I do X\", \"find a skill for X\", \"is there a skill that can...\", or is looking for functionality that might exist as an installable skill.",
      "path": "skills/find-skills",
      "version": "1.0.1"
    },
    {
      "name": "first-principles",
      "description": "Systems-level reasoning for high-stakes technical decisions. Use when choosing between architectures, evaluating trade-offs, or planning a non-mechanical refactor. For debugging a specific failure use diagnose; for clarifying an ambiguous WHAT use brainstorming.",
      "path": "skills/first-principles",
      "version": "2.0.1"
    },
    {
      "name": "frontend-design",
      "description": "Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.",
      "path": "skills/frontend-design",
      "version": "1.0.0",
      "license": "Complete terms in LICENSE.txt"
    },
    {
      "name": "gemini-imagen",
      "description": "Generate, edit, and compose images via Google's Gemini API. Use when the user mentions Gemini or wants multi-image composition and aspect-ratio control. Requires `GEMINI_API_KEY`.",
      "path": "skills/gemini-imagen",
      "version": "1.1.0"
    },
    {
      "name": "gh-commit-push-pr",
      "description": "Commit staged changes, push branch, and open a GitHub PR. Use when user asks to commit and push, create a PR, ship changes, send for review, or open a pull request. Triggers on phrases like 'commit and push', 'create a PR', 'open a pull request', 'send this for review', 'ship it', 'push and PR'.",
      "path": "skills/gh-commit-push-pr",
      "version": "1.0.2",
      "allowed-tools": "Bash(git checkout:*), Bash(git add:*), Bash(git status:*), Bash(git push:*), Bash(git commit:*), Bash(git diff:*), Bash(git log:*), Bash(git branch:*), Bash(git stash:*), Bash(git remote:*), Bash(gh auth status:*), Bash(gh repo view:*), Bash(gh pr create:*), Bash(gh pr view:*)"
    },
    {
      "name": "gpt-imagen",
      "description": "Generate or edit images via the OpenAI Image API (`gpt-image-2`). Use when the user asks to generate, edit, inpaint, mask, remove backgrounds, or create product shots, concept art, covers, or batch outputs using OpenAI. Requires `OPENAI_API_KEY`.",
      "path": "skills/gpt-imagen",
      "version": "1.1.1"
    },
    {
      "name": "handoff",
      "description": "Create session summaries for context preservation and handoff. Use when (1) conversation approaching context limits, (2) user requests session summary, (3) preparing handoff documentation to another agent or fresh instance, (4) archiving work for future reference.",
      "path": "skills/handoff",
      "version": "1.0.2"
    },
    {
      "name": "local-review",
      "description": "Perform local code reviews on workspace changes without posting to GitHub. Use for requests like /review, review this diff, audit staged changes, or check branch changes. Produces findings-first reports with severity, file line references, risks, and test gaps.",
      "path": "skills/local-review",
      "version": "1.1.4"
    },
    {
      "name": "loop-design",
      "description": "Design a reusable, verifiable autonomous loop on top of harness primitives like /loop and /goal. Use when setting up a recurring, unattended, or overnight agent loop, an automation or cron task, a /loop or /goal run, a Ralph-style while-true loop, or when deciding whether a task SHOULD be looped at all. Agent-agnostic across Claude Code, Codex, and CI. On-demand via /loop-design.",
      "path": "skills/loop-design",
      "version": "1.0.4"
    },
    {
      "name": "markdown-converter",
      "description": "Convert documents and files to Markdown using markitdown. Use when converting PDF, Word (.docx), PowerPoint (.pptx), Excel (.xlsx, .xls), HTML, CSV, JSON, XML, images (with EXIF/OCR), audio (with transcription), ZIP archives, YouTube URLs, or EPubs to Markdown format for LLM processing or text analysis.",
      "path": "skills/markdown-converter",
      "version": "1.0.0"
    },
    {
      "name": "nextjs-app-router",
      "description": "Next.js App Router expert guidance. Use when building, debugging, or architecting Next.js applications \u2014 routing, Server Components, Server Actions, layouts, data fetching, rendering strategies, metadata, image/font optimization, error handling, and hydration debugging. Triggers on tasks involving Next.js pages, app directory structure, RSC boundaries, or App Router migration.",
      "path": "skills/nextjs-app-router",
      "version": "1.0.1",
      "license": "MIT"
    },
    {
      "name": "obsidian-bases",
      "description": "Create and edit Obsidian Bases (.base files) with database-style views, filters, formulas, and summaries. Use when working with .base files, building table/card/dashboard views over a vault, or when the user mentions Bases, filters, or formulas in Obsidian.",
      "path": "skills/obsidian-bases",
      "version": "1.0.0"
    },
    {
      "name": "obsidian-canvas",
      "description": "Create and edit Obsidian Canvas files (.canvas) with nodes, edges, groups, and connections. Use when working with .canvas files, creating visual canvases, mind maps, flowcharts, or when the user mentions Canvas files in Obsidian.",
      "path": "skills/obsidian-canvas",
      "version": "1.0.0"
    },
    {
      "name": "obsidian-markdown",
      "description": "Create and edit Obsidian Flavored Markdown with wikilinks, embeds, callouts, properties, and other Obsidian-specific syntax. Use when working with .md files in Obsidian, or when the user mentions wikilinks, callouts, frontmatter, tags, embeds, or Obsidian notes.",
      "path": "skills/obsidian-markdown",
      "version": "1.0.0"
    },
    {
      "name": "playwright",
      "description": "Use when the task requires automating a real browser from the terminal (navigation, form filling, snapshots, screenshots, data extraction, UI-flow debugging) via `playwright-cli` or the bundled wrapper script.",
      "path": "skills/playwright",
      "version": "1.0.1"
    },
    {
      "name": "repo-hardening",
      "description": "Audit and harden a software repo against supply-chain and workflow risks. Use when a user asks to audit a repo, harden CI, pin GitHub Actions, freeze installs, reduce bot workflow risk, review lockfile/package-manager discipline, or create repo-local security investigation artifacts. Supports Node, Python, GitHub Actions, GitLab CI, and mixed-stack repos. On-demand via /repo-audit and /repo-harden.",
      "path": "skills/repo-hardening",
      "version": "1.0.2"
    },
    {
      "name": "research-architect",
      "description": "Engineer high-quality deep-research prompts and orchestrate their execution and verification. Use when the user wants to draft, improve, or critique a research prompt or brief; commission or plan a multi-source or multi-model research run; run research through external deep-research products (Claude/OpenAI/Gemini DR); or verify and score a research report that something else produced. Triggers on \"research prompt\", \"research brief\", \"commission research\", \"plan a research run\", \"verify this report\", \"research architect\". For a direct low-stakes lookup where the user just wants the answer, use deep-research instead.",
      "path": "skills/research-architect",
      "version": "2.3.0",
      "triggers": [
        "research prompt",
        "research brief",
        "research architect",
        "commission research",
        "plan a research run",
        "verify this research report"
      ]
    },
    {
      "name": "screenshot",
      "description": "Use when the user explicitly asks for a desktop or system screenshot (full screen, specific app or window, or a pixel region), or when tool-specific capture capabilities are unavailable and an OS-level capture is needed.",
      "path": "skills/screenshot",
      "version": "1.0.0"
    },
    {
      "name": "secure-code",
      "description": "Static analysis security scanning and architectural trifecta detection using semgrep. Use when reviewing code for security vulnerabilities, running SAST scans, checking for the lethal trifecta (private data + untrusted input + external comms co-occurrence), or when the user says 'scan', 'security scan', 'trifecta check', 'check for vulnerabilities', 'SAST', or 'secure this code'. On-demand via /scan and /trifecta-check commands.",
      "path": "skills/secure-code",
      "version": "1.0.2"
    },
    {
      "name": "session-retro",
      "description": "Update existing project reference docs with non-obvious learnings from the current session rather than creating new files. Use at session end, after solving tricky problems, or when new CLI commands/features were added. Triggers on \"/retro\", \"update docs with learnings\", \"save what we learned\", or proactively at session end.",
      "path": "skills/session-retro",
      "version": "1.0.3"
    },
    {
      "name": "skill-creator",
      "description": "Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends an AI agent's capabilities with specialized knowledge, workflows, or tool integrations.",
      "path": "skills/skill-creator",
      "version": "1.0.1",
      "license": "Complete terms in LICENSE.txt"
    },
    {
      "name": "skill-evals",
      "description": "Evaluate skill quality and routing reliability with deterministic checks. Use when creating/updating skills, validating trigger behavior (explicit/implicit/contextual/negative), applying SKILL.md contract checklists, or generating cross-skill compliance reports.",
      "path": "skills/skill-evals",
      "version": "1.5.0"
    },
    {
      "name": "skill-installer",
      "description": "Install skills into Codex or Claude Code skills directories from a curated list or a GitHub repo path. Use when a user asks to list installable skills, install a curated skill, or install a skill from another repo (including private repos).",
      "path": "skills/skill-installer",
      "version": "1.0.0"
    },
    {
      "name": "skill-standardizer",
      "description": "Use when skill copies drift across repositories or agent globals and you need canonicalization, drift auditing, and safe synchronization across local and global skills directories.",
      "path": "skills/skill-standardizer",
      "version": "1.3.2"
    },
    {
      "name": "template",
      "description": "Skill starter template with commented guidance for every contract section. Use when creating a new skill and you need a scaffold that passes strict contract validation.",
      "path": "skills/template",
      "version": "1.0.1"
    },
    {
      "name": "test-strategy",
      "description": "Guide agents to follow preferred testing methodology \u2014 red/green TDD, real dependencies over mocks, behavior-based tests, and effective-runtime authority-boundary probes. Use when writing tests, planning test coverage, deciding between TDD and test-after, correcting excessive mocking, or testing filesystem, credential, process, network, or remote-mutation permissions. Triggers on 'write tests', 'add test coverage', 'how should I test this', 'TDD', 'test strategy', 'test plan', 'test the permission boundary'.",
      "path": "skills/test-strategy",
      "version": "1.3.0"
    },
    {
      "name": "theme-factory",
      "description": "Toolkit for styling artifacts with a theme. Use when users ask to theme slides, docs, reports, or HTML pages using curated font/color systems or a generated custom theme.",
      "path": "skills/theme-factory",
      "version": "1.0.0",
      "license": "Complete terms in LICENSE.txt"
    },
    {
      "name": "type-design-review",
      "description": "Review the design of a new or changed type \u2014 does it make illegal states unrepresentable, enforce its invariants at construction, and encapsulate its internals? Rates encapsulation, invariant expression, invariant usefulness, and enforcement. Use when introducing a new type or data model, reviewing the types added in a PR, or refactoring a type for stronger guarantees. Not general diff review (local-review).",
      "path": "skills/type-design-review",
      "version": "1.0.0"
    },
    {
      "name": "vercel-composition-patterns",
      "description": "React composition patterns that scale. Use when refactoring components with boolean prop proliferation, building flexible component libraries, or designing reusable APIs. Triggers on tasks involving compound components, render props, context providers, or component architecture. Includes React 19 API changes.",
      "path": "skills/vercel-composition-patterns",
      "version": "1.0.1",
      "license": "MIT"
    },
    {
      "name": "vercel-deploy",
      "description": "Create and push new Vercel deployments. Use when the user wants to deploy, ship, go live, or create a preview \u2014 e.g. \"deploy my app\", \"push this live\", \"ship it\".",
      "path": "skills/vercel-deploy",
      "version": "1.0.1"
    },
    {
      "name": "vercel-preview-logs",
      "description": "Retrieve and diagnose Vercel build/runtime logs. Use when the user asks to check errors, inspect logs, debug a failed build, or correlate a PR/commit with log output from an existing environment.",
      "path": "skills/vercel-preview-logs",
      "version": "1.0.1"
    },
    {
      "name": "vercel-react-best-practices",
      "description": "React and Next.js performance optimization guidelines from Vercel Engineering. This skill should be used when writing, reviewing, or refactoring React/Next.js code to ensure optimal performance patterns. Triggers on tasks involving React components, Next.js pages, data fetching, bundle optimization, or performance improvements.",
      "path": "skills/vercel-react-best-practices",
      "version": "1.0.1",
      "license": "MIT"
    },
    {
      "name": "verify-before-complete",
      "description": "Guard against false completion claims when the cost of being wrong is high. Use when accepting delegated or subagent work, shipping high-risk changes (auth, migrations, infra, security, broad refactors), lacking fresh verification evidence (missing, stale, or conflicting), or being explicitly asked to confirm something is really done or audit a completion claim. Skip routine low-risk changes already covered by the repo's own checks \u2014 running those checks is enough.",
      "path": "skills/verify-before-complete",
      "version": "2.0.1"
    },
    {
      "name": "web-design-guidelines",
      "description": "Review UI code for Web Interface Guidelines compliance (accessibility, UX, code-level rules from Vercel WIG). Use when asked to \"review my UI\", \"check accessibility\", \"audit design\", \"review UX\", or \"check my site against best practices\". For visual taste / AI-slop audits use design-critique instead.",
      "path": "skills/web-design-guidelines",
      "version": "1.0.0"
    },
    {
      "name": "write-plan",
      "description": "Sequence the build: turn a settled target (a `write-spec` contract, a ticket, or a clear request) into an execution plan \u2014 task breakdown, files, ordered steps, seam selection, and verification commands. Use when WHAT is already decided and you need HOW: the file-level, dependency-ordered steps to implement it. If the target is not yet falsifiable, route back to `write-spec`.",
      "path": "skills/write-plan",
      "version": "2.5.0"
    },
    {
      "name": "write-spec",
      "description": "Define the target before building: write a falsifiable contract \u2014 problem, end-state, success criteria, evaluation \u2014 that states WHAT must be true, with no files or implementation steps. Use when you need to specify or align on what \"done\" means before sequencing work, or are handed a feature/change and must pin its acceptance criteria. Hand off to `write-plan` for the HOW.",
      "path": "skills/write-spec",
      "version": "2.2.0"
    }
  ]
}