# chiark-mcp

chiark-mcp discovers dependable AI agents for specific jobs, screens their live status, and filters by response delay.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · 432 installs
- Author: chiark-ai-dev
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/chiark-ai-dev/chiark-mcp
- Updated: 2026-03-30
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/mcp/chiark-mcp

## What it does

This tool links your environment to chiark.ai, an index tracking over 2,000 agents across nine public registries in A2A and MCP networks. The underlying service probes registered endpoints every half-hour to verify TLS status, response speeds, and protocol conformance.

It exposes five tools: `find_agent` for query-based routing with operational constraints, `check_agent_status` for real-time health checks, `get_agent_score` for rating breakdowns, `report_outcome` to submit routing feedback, and `get_ecosystem_stats` for aggregate network metrics. You can constrain discovery by minimum 30-day uptime, P95 latency thresholds, protocol flavor, or x402 payment requirements.

Services requiring credentials cap out at a score of 45 instead of 100 because full synthetic testing cannot run past authentication gates. You can route queries through a hosted HTTP endpoint or run a local Python binary.

## Requirements
- Needs: Python and uv (commands run via uvx)
- Remote server: `https://chiark-mcp--chiark-ai-dev.run.tools`

## Install

### Claude Code (native platform)

**Platform CLI: remote server**
```bash
claude mcp add --transport http chiark-mcp https://chiark-mcp--chiark-ai-dev.run.tools
```

### Codex

**Platform CLI: remote server**
```bash
codex mcp add chiark-mcp --url https://chiark-mcp--chiark-ai-dev.run.tools
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "chiark-mcp": {
            "url": "https://chiark-mcp--chiark-ai-dev.run.tools"
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "chiark-mcp": {
            "type": "http",
            "url": "https://chiark-mcp--chiark-ai-dev.run.tools"
        }
    }
}
```

### Gemini CLI

**Platform CLI: remote server**
```bash
gemini mcp add -t http chiark-mcp https://chiark-mcp--chiark-ai-dev.run.tools
```

## When to use it
- locate responsive MCP agents matching a precise task description
- confirm an agent endpoint is currently reachable before routing traffic
- exclude services that fail uptime thresholds or exceed acceptable latency
- retrieve diagnostic quality breakdowns across availability and protocol conformance

## Example prompts
- Use chiark-mcp to find agents handling web scraping with latency under 500 ms
- Check the live operational status of this agent ID before triggering the request
- Inspect the performance score breakdown for the chosen service

## Similar
- [Agent Discovery Board by SarnAI](https://affpapa.org/en/nevibecode/mcp/agent-discovery-board-by-sarnai.md) — MCP server
- [Structured Output MCP Agent](https://affpapa.org/en/nevibecode/mcp/structured-output-mcp-agent.md) — MCP server · 3042 installs
- [NEXUS Protocol](https://affpapa.org/en/nevibecode/mcp/nexus-protocol.md) — MCP server · 508 installs
- [agent-registry](https://affpapa.org/en/nevibecode/mcp/agent-registry.md) — MCP server · 2795 installs
- [google-agents-cli-publish](https://affpapa.org/en/nevibecode/skills/google-agents-cli-publish.md) — skill · 587200 installs
- [agent-protocol](https://affpapa.org/en/nevibecode/skills/borghei-agent-protocol.md) — skill · 104 installs

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
