Connect any MCP-compatible AI client to Agent Ledger. Log decisions, run risk scans, and generate compliance reports — directly from your AI assistant.
Ask Claude or Cursor to register agents, run risk classifications, log decisions, and generate audit reports — no manual API calls needed.
Register a new AI agent and receive an agent_id for subsequent compliance operations.
Classify an agent's EU AI Act risk level (minimal / limited / high / unacceptable) with actionable remediation steps.
Log an AI decision to the immutable audit trail. Required by Article 12 of the EU AI Act for high-risk systems.
Retrieve decision history for an agent. Filter by date, type, or confidence score for targeted audits.
Generate a full compliance report with decision statistics, risk indicators, and regulatory references.
Create API keys for programmatic access to Agent Ledger from your own applications and pipelines.
Retrieve the public compliance profile of an agent — shareable with clients and regulators.
List all registered agents in your account with status, risk class, and last activity.
Connect Claude Desktop or Cursor in under 5 minutes. No server setup required.
Register at Agent Ledger, create your first agent, and copy your al_live_... API key from the dashboard.
Edit your Claude Desktop config file:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"agentledger": {
"command": "npx",
"args": ["-y", "@agentledger/mcp"],
"env": {
"AGENTLEDGER_API_KEY":
"al_live_YOUR_KEY"
}
}
}
}
Restart Claude Desktop and try:
"Log a decision for my HR screening agent: it rejected application #4821 with confidence 0.87"
Claude will call Agent Ledger automatically and confirm the decision is logged.
One npm package. One environment variable. Compatible with any client that supports the Model Context Protocol.
Native MCP support in Claude Desktop. Restart the app after editing the config.
"command": "npx",
"args": ["-y", "@agentledger/mcp"],
"env": {
"AGENTLEDGER_API_KEY": "al_live_..."
}
Add Agent Ledger as an MCP server in Cursor settings under Features > MCP.
"agentledger": {
"command": "npx",
"args": ["-y", "@agentledger/mcp"],
"env": {
"AGENTLEDGER_API_KEY": "al_live_..."
}
}
Use the HTTP endpoint directly — no npm package needed for server-side agents.
POST /api/mcp
Authorization: Bearer al_live_...
Content-Type: application/json
{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": { ... }
}
Implement the stdio transport yourself or use the HTTP endpoint directly from any language.
AGENTLEDGER_API_KEY=al_live_... \ npx @agentledger/mcp
Pipe JSON-RPC over stdin/stdout.
Free tier includes 500 decisions/month. No credit card required. EU data residency guaranteed.
Create free account →