API Documentation

Query the Agentifact index programmatically. All endpoints return JSON with CORS enabled. Rate limited to 60 requests/minute per IP.

Base URL

https://agentifact.com

OpenAPI Spec

Machine-readable API specification: /openapi.yaml

Rate Limiting

60 requests per minute per IP address. Exceeding the limit returns 429 Too Many Requests with a Retry-After header in seconds.

GET/api/tools

List scored tools from the Agentifact index.

Query Parameters

categorystringFilter by AgentCategory enum (e.g. MCP_SERVER, FRAMEWORK)

approvalModestringFilter by approval mode (FULL_AUTO, REQUIRES_APPROVAL, HUMAN_IN_LOOP)

minScoreintegerMinimum composite trust score (0-100)

mcp"true"Filter to MCP-capable tools only

a2a"true"Filter to A2A-capable tools only

workflowstringFilter by workflow fit tag

limitintegerResults per request (1-100, default 20)

formatstringjson (default) or ndjson

Example

curl "https://agentifact.com/api/tools?category=MCP_SERVER&minScore=70&limit=10"

Response

{
  "meta": {
    "total": 10,
    "filters": { "category": "MCP_SERVER", "minScore": 70 },
    "schema_version": "1",
    "generated_at": "2026-03-26T00:00:00.000Z",
    "methodology": "/methodology"
  },
  "tools": [
    {
      "id": "langfuse",
      "name": "Langfuse",
      "category": "MONITORING",
      "scores": { "composite": 88, "trust": 85, ... },
      "protocols": { "mcp": true, "a2a": false, "rest": true },
      ...
    }
  ]
}

Scoring Methodology

Every tool is scored across 5 dimensions (0-100 each), weighted into a composite trust score:

Agent Readiness25%Protocol support, autonomy level, tool-calling readiness

Trust20%Vendor credibility, audit trail, governance controls

Interoperability20%Standards compliance, multi-protocol support

Security20%Auth, encryption, vulnerability disclosure

Documentation15%API docs, examples, maintenance signals

Full methodology: /methodology