Skip to main content
Glama
README.md
# agentiam-mcp

> Stdio MCP server wrapping all 18 [AgentIAM](https://achillesalpha.com) x402 endpoints — agent identity, safety, and intelligence for autonomous agents on Base Mainnet.

## What this is

`agentiam-mcp` exposes the AgentIAM HTTP catalog as a Model Context Protocol (MCP) server. Any MCP-capable client (Claude Desktop, Cursor, VS Code, LangChain, etc.) can call the 18 endpoints as tools.

Endpoints:
- **Safety pillars** — `noleak`, `memguard`, `riskoracle`, `secureexec`, `flowcore`
- **Audit / policy** — `audit`, `validate`, `risk-check`
- **Intelligence** — `research`, `intelligence-report`
- **DELPHI knowledge graph** — `delphi`, `delphi.entity`, `delphi.query`, `delphi.timeline`, `delphi.contradictions`
- **Signals** — `latest-signals`, `signal-query`, `publish-signal`

Pricing: $0.001–$0.05 USDC per call, settled on Base via x402.

## Status

**v0.1.0 — discovery mode.** All 18 tools are registered. Tool calls return the AgentIAM endpoint's decoded x402 challenge (resource URL, accepts array, bazaar extension, schema), so MCP clients can browse the catalog without holding USDC. Wallet-signed payment flow (per-call USDC settlement) ships in v0.2.

## Install

```bash
git clone https://github.com/achilliesbot/agentiam-mcp.git
cd agentiam-mcp
npm install
```

## Run (standalone)

```bash
node src/index.mjs
# stdio MCP server — speaks JSON-RPC over stdin/stdout
```

## Configure in Claude Desktop / Cursor / VS Code

Add to your MCP config (`claude_desktop_config.json` on macOS at
`~/Library/Application Support/Claude/claude_desktop_config.json`):

```json
{
  "mcpServers": {
    "agentiam": {
      "command": "node",
      "args": ["/absolute/path/to/agentiam-mcp/src/index.mjs"]
    }
  }
}
```

Restart the client. The 18 AgentIAM tools (`noleak`, `memguard`, `riskoracle`,
`secureexec`, `flowcore`, `audit`, `validate`, `risk_check`, `research`,
`intelligence_report`, `delphi`, `delphi_entity`, `delphi_query`,
`delphi_timeline`, `delphi_contradictions`, `latest_signals`, `signal_query`,
`publish_signal`) appear in the tool picker.

## Run via Docker

```bash
docker build -t agentiam-mcp .
docker run -i agentiam-mcp
```

## Smoke test

```bash
( echo '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"smoke","version":"0.1"}}}'
  sleep 0.3
  echo '{"jsonrpc":"2.0","method":"notifications/initialized"}'
  sleep 0.3
  echo '{"jsonrpc":"2.0","id":2,"method":"tools/list"}'
  sleep 1
) | node src/index.mjs
```

You should see a JSON-RPC response listing 18 tools.

## Discovery references

- AgentIAM x402 manifest: https://achillesalpha.com/.well-known/x402
- AgentIAM MCP discovery: https://achillesalpha.com/.well-known/mcp.json
- MCP Registry: `io.github.achilliesbot/ep-agentiam`
- x402scan: https://www.x402scan.com/server/de9dbadb-6475-43f2-a621-a805fb1c661e
- 402index: https://402index.io

## License

MIT.

TDQS

B3/5.0

Scored across 18 tools

Disambiguation4/5

Most tools have distinct purposes, especially the delphi_* cluster with specific query types. However, 'delphi' and 'latest_signals' both deal with real-time signals and could cause confusion, though descriptions differentiate them. Similarly, 'risk_check' and 'riskoracle' overlap in risk assessment but with different scopes. Overall, agents can likely select the correct tool with careful reading.

Naming Consistency3/5

Naming uses a mix of conventions: single words ('audit', 'research'), concatenated words ('flowcore', 'riskoracle'), and underscores ('delphi_entity', 'risk_check'). The delphi_* tools follow a consistent pattern, but the rest do not follow a uniform verb_noun or other clear pattern. This mixed style reduces predictability.

Tool Count4/5

18 tools is on the higher side but reasonable for the broad domain covering security audits, intelligence signals, risk analysis, and execution verification. Each tool serves a specific function, and the count does not feel excessive given the scope. However, a slight reduction could improve coherence.

Completeness4/5

The tool set covers most core workflows: signal querying, publishing, risk assessment, verification, and orchestration. Minor gaps exist, such as lacking signal update/delete tools and possibly entity management for the knowledge graph. Overall, the surface is largely complete for security and intelligence tasks.

Maintenance

ActivityInactive
ResponsivenessNo issues