Mode
Use all approved listings
Agents can discover only listings that pass the approval gate and have no unresolved reports.
Agent Exchange
Separate human browsing from agent access so only approved, report-free resources enter runtime discovery.
Why this page exists
A policy mode narrows what an agent can discover; it does not replace source review, local testing, or project-specific approval.
Mode
Agents can discover only listings that pass the approval gate and have no unresolved reports.
Mode
Teams can narrow access to named resources after reviewing source, license, outputs, and safety notes.
Mode
Teams can exclude specific resources from an approved set when project policy or risk posture changes.
Approval boundary
Catalog visibility means a human can inspect a listing; it is not the same as agent-facing access.
Agent-facing discovery depends on the manifest, approval state, and unresolved-report checks.
Open reports keep a listing out of agent-facing results until the concern is resolved.
Teams still review behavior in a low-risk workspace before using a resource in important work.
Access routes
Policy decisions stay tied to the catalog, compare table, manifests, and support routes that expose the evidence behind a listing.
Browse public listings and filter them by resource type.
Review selected listings side by side.
Read about review states, source information, and safe use.
Inspect approved discovery and manifest endpoints.
Get help with a source, review status, license, or report.
Learn how reusable resources can support Flow work inside Matrix.
Return to the Agent Exchange overview.
Approved catalog
Reusable Copilot prompts, instructions, chat modes, and customizations.
Configuration files that enhance Cursor AI editor rules and behavior.
Curated prompts for Cursor Rules, Cline, Windsurf, and GitHub Copilot.
Task management patterns for coordinating AI development agents.
Practical context-engineering examples and templates for agentic coding.
A broader context-engineering resource set for AI agents and coding workflows.
AGENTS.md-style rules inspired by software engineering books and patterns.
CLI for applying the same coding-agent rules across several assistants.
Utility CLI for synchronizing AI coding-agent rule files.
Zero-dependency rule converter for common AI coding-agent rule formats.
Curated directory of Model Context Protocol servers and integrations.
Official Python SDK for building Model Context Protocol clients and servers.
GitHub's MCP server for repository, issue, pull request, and code workflows.
MCP server for browser automation and web app verification with Playwright.
MCP server that gives coding agents current library documentation context.
MCP server for web crawling and structured extraction workflows.
MCP server for Exa search and research integrations.
Supabase MCP server for database and project workflows.
MCP server for Cloudflare account and developer platform workflows.
Open-source MCP servers for AWS developer and cloud workflows.
Docker MCP CLI plugin and gateway for managing MCP server access.
MCP server for JetBrains IDEs including IntelliJ, PyCharm, and WebStorm.
MCP server for browser automation through Browserbase and Stagehand.
Framework for managing multi-language pre-commit hooks.
Maintained collection of useful hooks for common repository hygiene checks.
Secret detection scanner that can run locally, in hooks, or in CI.
Yelp's secret-detection tool for baselined repository scans.
Fast source-code spell checker suitable for local hooks and CI.
pre-commit integration for the Ruff Python linter and formatter.
Automated code review runner that integrates with language-specific analyzers.
Fast Git hooks manager for multi-language repositories.
Git hook tooling for JavaScript and TypeScript projects.
Runs formatters and linters against staged files.
Commit-message linting for conventional commit workflows.
Git hook utility for preventing known credential patterns from being committed.
Command-line Markdown linter suitable for docs hooks and CI.
Python SDK for building agent workflows with tools, handoffs, and tracing.
JavaScript and TypeScript SDK for OpenAI agent workflows.
Educational framework exploring lightweight multi-agent orchestration.
Open-source coding agent that runs in the terminal.
Framework for durable, stateful, multi-step agent workflows.
Framework for LLM applications with retrieval, tools, and agent primitives.
Microsoft SDK for composing agents, plugins, memory, and planners.
Framework for role-based AI agent teams and workflows.
Data framework for connecting private or domain data to LLM apps.
Deepset framework for production-ready search, RAG, and agent pipelines.
Framework for programming and optimizing language-model pipelines.
Hugging Face library for simple code-agent and tool-calling workflows.
Agent framework from Pydantic focused on typed outputs and Python ergonomics.
Agent framework and research platform for communicative multi-agent systems.
Library for connecting AI agents to browser automation tasks.
Browserbase project for AI-assisted browser automation with Playwright.
AI pair-programming tool that edits code in a local git repository.