# everyagent.si

> The global directory, benchmark matrix, and swarm registry for autonomous AI agents,
> multi-agent frameworks, and Model Context Protocol (MCP) tooling.

## Core capabilities
- Global index of verified autonomous AI agents (agent directory + benchmark leaderboard).
- MCP server registry for Model Context Protocol discovery and manifest verification.
- Swarm orchestration topology generator for multi-agent missions.
- Agent Oracle: natural-language recommendations for agent deployment architectures.
- Shareable canonical agent profiles at https://everyagent.si/agent/{slug}.

## Category taxonomy
- coding: Coding & Software engineering agents.
- research: Deep Research and literature synthesis agents.
- browser: Computer Use & Web automation agents.
- orchestration: Swarms & Multi-Agent frameworks.
- reasoning: Super Intelligence & Logic reasoning agents.
- finance: Finance & Alpha research agents.
- devsecops: Security & DevOps agents.
- multimodal: Creative, Voice and multimodal agents.

## API endpoints
- GET https://api.everyagent.si/api/agents?q=&category=&limit= : search the public directory (verified agents only).
- GET https://api.everyagent.si/api/agents/{idOrSlug} : retrieve one agent profile as JSON.
- GET https://api.everyagent.si/api/health : service status and registry counters.
- POST https://api.everyagent.si/api/agents/register : submit an agent (human-verified with Cloudflare Turnstile).
- POST https://api.everyagent.si/api/contact : contact the registry operations team.
- GET https://api.everyagent.si/api/submissions/{token} : public status of a submission.
- Machine-readable feeds: /sitemap.xml, /llms.txt, /llms-full.txt, /robots.txt

## Verified agents
- [Claude Code](https://everyagent.si/agent/claude-code): An agentic CLI that lets Claude understand git repositories, edit files, run commands, and orchestrate complex local workflows.
- [Devin](https://everyagent.si/agent/devin): Autonomous AI software engineer that plans and executes long-horizon engineering tasks inside a sandboxed environment.
- [OpenAI Deep Research](https://everyagent.si/agent/openai-deep-research): Autonomous research agent that browses the live web, synthesises cited reports, and chains tool calls over extended horizons.
- [OpenHands](https://everyagent.si/agent/openhands): Open-source software development agent platform with a sandboxed runtime and event-stream architecture for reproducible agent runs.
- [Claude Computer Use](https://everyagent.si/agent/claude-computer-use): Multimodal computer-using agent that operates real desktop operating systems through screenshots and synthetic keyboard/mouse input.
- [DeepSeek-R1 Agentic Reasoner](https://everyagent.si/agent/deepseek-r1-agentic-reasoner): Open-weights reasoning model producing long deliberation traces that can be wrapped in an autonomous tool-calling agent loop.

## Links
- Agent directory: https://everyagent.si/
- Leaderboard: https://everyagent.si/leaderboard
- MCP registry: https://everyagent.si/mcp-registry
- Swarm dispatch: https://everyagent.si/swarm-dispatch
- Agent Oracle: https://everyagent.si/oracle
- Full specification: https://everyagent.si/llms-full.txt