The verified tool registry

Open-source, code-first toolkit for building, evaluating, and deploying sophisticated multi-agent AI systems, optimized for Gemini but model- and deployment-agnostic.

A Ralph-style loop that turns open GitHub issues into pull requests, one task per iteration, until none remain.

Anthropic's public repository of Agent Skills: self-contained folders with a SKILL.md that Claude loads dynamically to perform specialized tasks like document creation and MCP server generation.

Vercel's official collection of agent skills for AI coding agents, covering React/Next.js performance, AI SDK usage, web UI guidelines, and Vercel deploys.

Cloudflare's SDK for building and deploying stateful, persistent AI agents on Workers and Durable Objects, with built-in state, scheduling, WebSockets, MCP, and React hooks.

Provider-agnostic TypeScript toolkit for building AI-powered applications and agents across Next.js, React, Svelte, Vue, and Node.js.

Create algorithmic and generative art with p5.js using seeded randomness and interactive parameter exploration — flow fields, particle systems, and more.

Systematic requirements discovery through focused multi-round questioning before any code gets written — asking only what the code can't reveal.

Programming framework for agentic AI and multi-agent conversation patterns (now in maintenance mode).

Test-first loop: write a failing test, implement the minimum code to pass, refactor, and repeat.

GitHub's curated marketplace of custom agents, skills, instructions, hooks, and plugins to extend GitHub Copilot.

The unified command line interface for managing and automating all AWS services from your terminal.

An official open-source suite of Model Context Protocol servers connecting AI assistants to AWS services, docs, and infrastructure.

A command line tool to build, test, debug, and deploy serverless applications using the AWS Serverless Application Model.

Build with Azure AI services — AI Search (vector/hybrid/semantic), Speech, OpenAI, and Document Intelligence (OCR).

Azure cost management — query and forecast spending, find orphaned resources, and optimize to reduce waste.

Execute Azure deployments for already-prepared apps with existing deployment plans and infrastructure files.

Debug Azure production issues with AppLens, Azure Monitor, and resource health — safe triage for App Service, containers, and more.

Plan, create, and configure production-ready Azure Kubernetes Service (AKS) clusters — SKU selection, networking, and security.

MCP server connecting AI agents to 40+ Azure service areas for resource management and queries.

Find the least-privilege Azure RBAC role for an identity and generate the CLI commands and Bicep to assign it.

Microsoft's Azure capability plugin: curated Azure skills plus the Azure MCP Server (200+ tools across 40+ services) and Foundry MCP, for real Azure work from coding agents.

Work with Azure Storage — Blob, File, Queue, Table, and Data Lake — including access-tier guidance (hot/cool/cold/archive).

Drive a GitHub PR to a fresh approval, green CI, and a mergeable state — verifying every review finding against the code, fixing the real ones, and pushing back with evidence on the rest.

Green tests are not a running app — build the production image, boot it against a fresh throwaway database, probe the real routes and contracts, fix, and re-boot until everything answers.

Apply Anthropic's official brand colors and typography to any artifact that benefits from a consistent company look-and-feel.

Run the production build, fix compile and bundling errors, and loop until the build succeeds.

Create beautiful static visual art as .png and .pdf — posters, designs, and artwork — guided by a design philosophy.

Ultra-compressed communication mode that cuts token usage ~75%.

After shipping a feature, loop until CHANGELOG.md has a user-facing entry and conventional commits are reflected.

Write the success criteria and load-bearing assumptions first, complete exactly one step per pass while restating done / verified / left, and loop until every criterion is checked off and backed by evidence.

TypeScript framework for building custom AI agents powered by Claude Code, supporting one-off queries, interactive client sessions, custom tools, and hooks.

Python framework for building custom AI agents on top of Claude Code, with query() for one-off tasks and ClaudeSDKClient for interactive sessions, custom tools, and hooks.

Reference for building with the Claude API / Anthropic SDK: models, params, streaming, tool use, MCP, caching, and token counting.

Agentic coding tool that lives in your terminal, understands your codebase, and executes routine tasks, explains code, and handles git workflows via natural language.

Official Go-based CLI for the Claude Developer Platform / Claude API, distinct from the agentic Claude Code tool.

Anthropic-managed directory of high-quality Claude Code plugins, including internal Anthropic plugins and vetted third-party submissions, that extend Claude Code with commands, agents, and MCP servers.

A token-efficient remote MCP server exposing the entire Cloudflare API (~2,500 endpoints) via a Code Mode search/execute pattern at mcp.cloudflare.com.

A suite of official remote MCP servers (observability, bindings, builds, browser rendering, Radar, AI Gateway, and more) that let MCP clients manage Cloudflare via natural language.

Official collection of Agent Skills, slash commands, and bundled MCP servers that teach coding agents (Claude Code, Cursor, Codex, etc.) how to build on Cloudflare.

Lightweight coding agent from OpenAI that runs locally in your terminal, usable with a ChatGPT plan or API key.

GitHub Action that runs Codex inside CI/CD workflows to review PRs or apply patches, with configurable sandbox privileges.

Claude Code plugin that lets you run Codex from within Claude Code to review code or delegate background tasks using your local Codex auth.

React composition patterns that scale — use when refactoring components toward more composable, maintainable structures.

Align new code to the codebase's own idioms — sample its neighbors, fix one mismatch per pass, and when two patterns conflict pick the more recent or better-tested one rather than averaging them.

Add focused tests until coverage meets your threshold (e.g. 80%), without changing production behavior unnecessarily.

Official scaffolding CLI that bootstraps and deploys new Workers and framework-based projects using Cloudflare-maintained templates, and installs Wrangler.

After implementation, run a cleanup pass: remove debug code, tighten naming, delete dead branches, and align with project conventions.

Collapse a menu of options into one committed recommendation — research each candidate with real evidence instead of priors, score against explicit criteria, and don't stop until a single pick stands with a named runner-up.

Deploy applications and websites to Vercel — production or preview deployments — straight from a coding agent.

On an interval, hit health and smoke endpoints after a deploy until all checks return healthy responses.

A disciplined diagnosis loop for hard bugs and performance regressions.

Guide a structured workflow for co-authoring documentation, proposals, technical specs, and decision docs through iterative refinement.

Open-source 'docker agent' builder and runtime for declaring, running, and sharing multi-agent AI teams via YAML, with MCP tools and any LLM provider (formerly named cagent).

The official Docker command-line interface for building, running, and managing containers, images, networks, and volumes against the Docker daemon.

Docker CLI plugin to define and run multi-container applications from a single declarative YAML file with one command.

Open-source 'docker mcp' CLI plugin and gateway that orchestrates Model Context Protocol servers as isolated containers, with secrets, OAuth, and access control, powering Docker Desktop's MCP Toolkit and catalog.

The 'docker model' CLI plugin for pulling, running, packaging, and serving local LLMs and AI models as OCI artifacts with an OpenAI-compatible API.

The 'docker scout' CLI plugin for software supply chain security: generating SBOMs, matching components against vulnerability databases, and evaluating image policy compliance.

Treat every operational claim in your docs and runbooks as a hypothesis — verify each against the live system, fix what drifted, and loop until no claim fails verification.

Create, read, edit, and manipulate Word (.docx) documents, including tables of contents, headings, and professional formatting.

Extract a shared unit only where the same logic appears three or more times — and inline back any abstraction used only once or twice.

A harness-native operator system — skills, instincts, memory, and security — that tunes Claude Code, Codex, OpenCode, and Cursor for research-first development.

Unofficial CLI for the Fatture in Cloud API v2 — JSON-first, jq-pipeable, agent-friendly.

Official command-line interface for Firebase to initialize, test, manage, and deploy Firebase projects, Cloud Functions, hosting, and AI/Genkit features.

Fetch the latest failed CI run, reproduce the failure locally, fix it, and repeat until checks pass.

Run the formatter, fix any remaining style issues, and loop until the codebase formats cleanly with no diffs.

Create distinctive, production-grade frontend interfaces with high design quality — components, pages, and full applications.

Open-source AI agent that brings Gemini directly into your terminal, using a ReAct loop with built-in tools and MCP server support.

Open-source framework (with CLI and local dev UI) by Google's Firebase for building AI-powered and agentic apps in JavaScript, Go, and Python with a unified model interface, flows, and RAG.

GitHub's official command-line tool for working with pull requests, issues, releases, and the GitHub API from the terminal.

Brings the Copilot coding agent to the terminal, letting you build, debug, and understand code via natural language.

GitHub's official Model Context Protocol server that connects AI agents to GitHub repos, issues, pull requests, and workflows.

Official Google Cloud command-line interface (part of the Cloud SDK, including gcloud, gsutil, and bq) for managing Google Cloud resources, AI/Vertex services, and applications.

Get interviewed relentlessly about a plan until it's fully resolved.

Stress-test a plan against your domain model and update the docs inline.

Audit a doc, PR description, or answer for unsourced assertions and replace each hedge with a cited artifact — file:line, command output, or URL — until nothing rests on inference.

Garry Tan's exact Claude Code setup — 23 specialist agents and 8 power tools as slash-command skills that turn Claude Code into a virtual engineering team.

Compact the conversation into a handoff document for another agent.

The official Python client and `hf` command-line tool for interacting with the Hugging Face Hub (auth, downloading/uploading models, datasets, and Spaces).

Official MCP server that connects MCP-compatible assistants to the Hugging Face Hub for searching models, datasets, Spaces, papers, and docs, plus running Gradio community tools.

Find architectural deepening and refactoring opportunities in a codebase.

When implementation claims done, a separate verifier pass runs build, lint, and tests with no access to the implementer's rationale.

Mutate your own production code on purpose, prove a test fails for each change, and loop until no surviving mutant slips past the suite.

Write internal communications — status reports, leadership updates, newsletters — in the formats a company prefers.

Before a large batch job, surface the few decisions only the owner can make, lock them in one up-front interview, then run the whole batch unattended — looping on blockers, not on questions.

Command-line companion and TUI for Jules, Google's asynchronous cloud coding agent, to create and manage agent sessions that write tests, fix bugs, and open pull requests.

AWS's agentic command-line tool for building software through natural-language chat in your terminal, the successor to the Amazon Q Developer CLI.

End-to-end robotics toolkit with CLI commands (`lerobot-train`, `lerobot-eval`, `lerobot-info`) for training, evaluating, and running real-world robot policies on PyTorch.

Run lint and typecheck, fix reported issues, and loop until the codebase is clean.

MCP server exposing MarkItDown's file-to-Markdown conversion (PDF, Office, images, audio) to agents.

Guide for building high-quality MCP servers in Python (FastMCP) or Node/TypeScript (MCP SDK) to integrate external APIs and services.

Curated reference MCP server implementations (Filesystem, Fetch, Git, Memory, Sequential Thinking, Everything) maintained by the Anthropic-managed Model Context Protocol project to demonstrate MCP features and SDK usage.

Open-source MCP server (formerly Gen AI Toolbox for Databases) that securely connects AI agents and MCP clients to databases like AlloyDB, Spanner, Cloud SQL, BigQuery, and self-managed Postgres/MySQL.

CLI that generates versioned, code-owned AI SDK tool stubs from any MCP server to cut token cost and reduce prompt-injection risk.

After merging, loop until production actually reflects the merge — pipeline green, the new commit serving, fresh routes answering, and schema plus data applied.

Multi-language framework for building production AI agents and multi-agent workflows; successor to Semantic Kernel and AutoGen.

Deploy, evaluate, fine-tune, and manage Microsoft Foundry agents end-to-end — builds, hosted agents, batch/continuous eval, and prompt optimization.

Run database migrations, fix schema issues, and loop until migrations apply cleanly in dev.

An open-source ML engineer agent that reads papers, trains models, and ships them, routing tool calls across the Hugging Face Hub, Jobs, and sandboxed execution.

A Claude Code plugin that lets you invoke Hugging Face's ml-intern agent to fine-tune, evaluate, and ship models directly from inside Claude Code.

A markdown-and-bash protocol that turns a well-scoped repo into a supervised multi-agent coding loop where a Beacon ranks the next work, you approve it, and path-owning agents build in isolated git worktrees.

Official command-line tool for developers and coding agents to authenticate, make Notion API requests, upload files, and build/deploy Notion Workers from the terminal.

Official Notion MCP server giving AI tools secure access to a Notion workspace; a hosted remote variant is offered at https://mcp.notion.com/mcp via OAuth.

Official Cursor plugin packaging Notion Skills with Notion's hosted MCP server (https://mcp.notion.com/mcp) for one-click workspace access inside Cursor.

Official collection of agent skills for working with Notion, including a notion-cli skill for installing and using the ntn CLI via coding agents.

Official Claude Code plugin that bundles Notion Skills, the Notion MCP server, and slash commands for one-click connection of Claude Code to a Notion workspace.

A multi-model coding-agent harness that installs into OpenCode and the OpenAI Codex CLI.

JavaScript/TypeScript framework for building multi-agent and voice agent workflows with handoffs, guardrails, and tracing.

Lightweight, provider-agnostic framework for building multi-agent workflows with tools, handoffs, guardrails, sessions, and tracing.

Official, type-annotated Python client library for the OpenAI REST API, with synchronous and asynchronous clients for the Responses, Chat Completions, and Realtime APIs.

Work with PDF files: extract text and tables, merge and split, rotate, watermark, fill forms, encrypt, and create new PDFs.

A self-extensible coding-agent harness and AI agent toolkit: an interactive coding agent CLI with a unified multi-provider LLM API, TUI, and web UI libraries.

MCP server that gives agents structured browser automation via Playwright's accessibility tree.

Teach your agent the positronick.com registry: find, inspect and install souls and agent tooling through the MCP server or the CLI.

Official Language Server and CLI toolchain for Postgres, providing autocompletion, type checking, linting, and syntax error detection built on Postgres' own libpg_query parser.

Official MCP server that lets LLMs perform CRUD operations over a Supabase/PostgREST REST API and convert SQL queries to PostgREST syntax, ideal for connecting end users to your app's data.

Create, read, and edit PowerPoint (.pptx) decks — slide decks, pitch decks, and presentations — or extract their content.

Review your own diff like a senior reviewer, fix issues, and repeat for three passes before opening the PR.

Build a throwaway prototype to flesh out a design before committing.

Repeatedly relaunch a fresh coding-agent session against a prompt until the implementation plan is complete.

React and Next.js performance optimization guidelines from Vercel Engineering, for writing, reviewing, or refactoring components.

React Native and Expo best practices for building performant mobile apps with coding agents.

Implement smooth, native-feeling animations with React's View Transition API — page transitions, route animations, and CSS view-transition pseudo-elements.

Before editing a symbol, map every caller, re-export, test, and shared util it touches — iterate until the blast radius is fully known, then write.

Attack your own just-built feature's trust assumptions — spoofable identifiers, client-settable fields, lookalike handles, reclamation attacks — fix each break and re-attack until every attack fails.

Distill a coding session's hard-won lessons — gotchas, patterns, integration quirks — into the project's knowledge base, with explicit approval before anything is written.

A command-line tool that flattens any git repo into a single static HTML page for fast human skimming or for copying the whole codebase into an LLM.

After every schema edit, regenerate migrations and prove there is no drift — loop until a fresh generate produces no new files and leaves the working tree clean.

Model-agnostic SDK for orchestrating, building, and deploying AI agents and multi-agent systems.

The newer TypeScript Sentry CLI built for developers and AI agents, with Seer-powered root cause analysis (sentry issue explain/plan), DSN auto-detection, and JSON output.

Official Model Context Protocol server that connects AI coding assistants to Sentry's API for searching issues, retrieving events, and debugging production errors; hosted remotely at mcp.sentry.dev.

The long-standing Rust-based command-line utility for Sentry, used to upload debug symbols and source maps, manage releases, and view issues.

Official Sentry plugin for Claude Code and Cursor that bundles agent skills, the /seer slash command, and auto-configured Sentry MCP to set up Sentry and fix production issues.

Open-format Agent Skills used internally by the Sentry team for development, including code review, commit, branch-naming, and security/performance review skills that work across Claude Code, Cursor, Cline, and Copilot.

After significant work, distill the non-obvious lessons — gotchas, environment quirks, corrections you received — into durable notes, looping until a fresh review pass surfaces nothing unrecorded.

One-time per-repo setup that teaches the engineering skills your conventions.

Take a finished change from working tree to a green, mergeable PR — one command chaining simplify, verify, code-review, smart-commit, branch hygiene, PR, and babysit-to-green.

Implement on a branch, run tests, push, open a PR, and loop until checks pass and the PR is ready to merge.

Hunt down swallowed errors, empty catch blocks, ignored return codes, silent fallbacks, and skipped tests, then make each one fail loud or carry an explicit comment justifying the suppression.

Create, modify, and improve Agent Skills, and measure their performance with evals, benchmarks, and variance analysis.

Open agent-skills CLI to install and manage reusable skill packages across Claude Code, Cursor, Codex, OpenCode, and 60+ coding agents.

Official catalog of Agent Skills (folders of instructions, scripts, and resources) that Codex and other AI agents can discover and use.

Create animated GIFs optimized for Slack, with constraints, validation tools, and animation concepts.

Cluster uncommitted changes into logical groups and commit each with a clean conventional commit message — plus a Keep-a-Changelog update pass.

A barebones Python library for building agents that think and act in code, with the `smolagent` and `webagent` CLIs for running multi-step and web-browsing agents.

GitHub's toolkit and Specify CLI for Spec-Driven Development, generating executable specs and slash-command workflows for AI coding agents.

Implement from a written spec.md checklist — each iteration completes one unchecked requirement with verification.

When your sync pipeline only upserts, removals never propagate — loop until the live store's rows exactly match the curated source files, pruning orphans deliberately.

An AWS-built, model-driven open-source SDK for building and running AI agents in a few lines of code, used in production by Kiro, Amazon Q, and AWS Glue.

Official Python and TypeScript library that integrates Stripe APIs into agent frameworks (OpenAI Agents SDK, LangChain, CrewAI, Vercel AI SDK) and MCP via function calling.

Official command-line tool to build, test, and manage a Stripe integration — trigger and forward webhooks, tail API logs, and CRUD API objects from the terminal.

Official CLI and local MCP server that issues secure single-use Link payment credentials so AI agents can make approved purchases on your behalf without exposing real card details.

Official Model Context Protocol server exposing Stripe payments, billing, and docs/knowledge-base tools to AI agents, runnable locally via npx or as a hosted remote server at https://mcp.stripe.com.

Official Agent Skills that teach AI coding agents (Claude Code, Copilot, Cursor, and 18+ others) how to build correctly on Supabase, covering Database, Auth, Edge Functions, Storage, Realtime, and Postgres best practices.

Official CLI to run the full Supabase stack locally, manage Postgres migrations, deploy edge functions, run backups, and generate types from your schema.

Official MCP server that connects AI assistants like Cursor, Claude, and Windsurf to your Supabase projects to manage tables, query data, deploy edge functions, and handle config.

An agentic skills framework and complete software-development methodology that makes your coding agent plan, spec, and verify before it writes code.

Shrink an over-broad diff back to the minimum that satisfies the goal — every changed line load-bearing, no drive-by edits — re-running tests after each cut.

Engineering-preview orchestrator that turns project work into isolated, autonomous Codex implementation runs by monitoring a board and spawning agents.

Test-driven development with a strict red-green-refactor loop.

Learn a new skill or concept, grounded in your workspace.

Run your test suite, fix the smallest root cause on failure, and repeat until every test passes.

An agent skill that authors Lottie (Bodymovin) JSON animations rendered live in an official Skia/Skottie player — for creating, editing, or fixing animations with coding agents.

Style artifacts — slides, docs, HTML landing pages — with one of 10 preset themes or a theme generated on the fly.

Break a plan or PRD into independently-grabbable tracker issues.

Turn the current conversation into a PRD on your issue tracker.

Triage incoming issues through a role-driven state machine.

Strongly-typed superset of JavaScript that compiles to plain JS via the tsc command-line compiler.

Official TypeScript SDK for the v0 Platform API to programmatically create AI-generated chats, projects, and deployments; ships with a create-v0-sdk-app scaffolder.

Official command-line interface for deploying and managing Vercel projects, env vars, domains, logs, and deployments from the terminal.

Deploy and manage Vercel projects with token-based (non-interactive) auth — deploys, setup, and environment variables via the Vercel CLI.

Vercel's official remote MCP server with OAuth, letting AI tools securely search docs and manage Vercel projects, deployments, and logs.

Vercel cost and performance optimization for deployed Next.js, SvelteKit, Nuxt, and Astro apps — metric-backed, evidence-first recommendations.

For every candidate item in a catalog or dataset, independently re-verify the claim and mechanically confirm its source URL resolves — include it with evidence or drop it with a written reason.

Build elaborate multi-component claude.ai HTML artifacts with React, Tailwind CSS, and shadcn/ui — for artifacts needing state, routing, or components.

Review UI code for Web Interface Guidelines compliance — accessibility, UX, and design best-practice audits.

Interact with and test local web apps via Playwright: verify frontend behavior, debug UI, capture screenshots, and view browser logs.

The official command-line interface for the Cloudflare Developer Platform, used to build, test, configure, and deploy Workers and related services.

Create new agent skills with proper structure and progressive disclosure.

Review docs and prose for Writing Guidelines compliance — voice, tone, and style audits against a writing handbook.

Open, read, edit, and create spreadsheets (.xlsx, .xlsm, .csv, .tsv) — formulas, formatting, charts, and data cleaning.

A local MCP server that turns the X API's OpenAPI spec into callable tools (streaming and webhook routes excluded), served over HTTP at /mcp.

A curl-like CLI for the X API that handles OAuth 1.0a and OAuth 2.0 (PKCE, multi-account) and supports streaming, media uploads, and webhooks.

Run lint, typecheck, tests, and build as one chained gauntlet under a 'a warning FAILS' bar, and loop until the whole chain exits clean with the word 'warning' nowhere in its output.

Get broader, higher-level context on unfamiliar code.