API Documentation

Everything you need to integrate AgentSource skills into your AI agent.

Base URL: https://api.agentsource.co

Quick Start

Get from zero to calling a skill in 3 steps:

1Get an API Key

2Find a Skill

bash

curl https://api.agentsource.co/v1/skills/search?q=scrape+web \
  -H "Authorization: Bearer YOUR_API_KEY"

3Call the Skill via MCP

bash

curl -X POST https://proxy.agentsource.co/v1/mcp/web-scraper \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "scrape_url",
      "arguments": { "url": "https://example.com" }
    }
  }'

Authentication

AgentSource uses API keys for agent-to-proxy authentication and JWT tokens for the human dashboard.

Sign Up

bash

curl -X POST https://api.agentsource.co/v1/auth/signup \
  -H "Content-Type: application/json" \
  -d '{"email": "agent@example.com", "password": "secure123", "name": "My Agent"}'

Returns a JWT token. Use it to create API keys via the dashboard API.

Create API Key

bash

curl -X POST https://api.agentsource.co/v1/dashboard/api-keys \
  -H "Authorization: Bearer JWT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "production-agent"}'

The API key is returned once — store it securely. Use it as X-API-Key header for MCP proxy calls.

Search Skills

Vector Search

Semantic search powered by embeddings + trigram fallback.

http

GET /v1/skills/search?q=convert+markdown&limit=10&category=data

Response:
{
  "data": [
    {
      "id": "abc-123",
      "slug": "markdown-converter",
      "name": "Markdown Converter",
      "short_description": "Convert between Markdown, HTML, and plain text",
      "trust": { "score": 88, "total_calls": 0 },
      "tool_count": 4
    }
  ]
}

List Skills

Browse with filters and sorting.

http

GET /v1/skills?category=dev-tools&sort=trust&per_page=20

Query params:
  category   - Filter by category slug
  pricing    - "free" | "per_call"
  min_trust  - Minimum trust score (0-100)
  sort       - "popular" | "newest" | "trust" | "name"
  per_page   - Results per page (default: 20)

Skill Detail

http

GET /v1/skills/:slugOrId

Returns full detail: description, tools with input schemas,
trust metrics, versions, screening status, author info.

Connect to a Skill

Before calling a skill through the proxy, register a connection. This enables metering and quota tracking.

bash

curl -X POST https://api.agentsource.co/v1/agent/connect \
  -H "Authorization: Bearer JWT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"skill_id": "abc-123"}'

# Response:
{
  "data": {
    "connection_id": "conn-456",
    "proxy_endpoint": "https://proxy.agentsource.co/v1/mcp/markdown-converter"
  }
}

MCP Proxy

The proxy gateway handles auth, rate limiting, metering, and forwards MCP JSON-RPC requests to skill servers.

Endpoint: POST https://proxy.agentsource.co/v1/mcp/:skillSlug

List Available Tools

json

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/list",
  "params": {}
}

Call a Tool

json

{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "tools/call",
  "params": {
    "name": "scrape_url",
    "arguments": {
      "url": "https://httpbin.org/html",
      "format": "markdown"
    }
  }
}

Required Headers

Header	Value
Content-Type	application/json
Accept	application/json, text/event-stream
X-API-Key	Your API key

All Endpoints

Method	Path	Description
POST	/v1/auth/signup	Create account
POST	/v1/auth/login	Get JWT token
GET	/v1/skills	List skills with filters
GET	/v1/skills/search	Vector search skills
GET	/v1/skills/:id	Skill detail
GET	/v1/categories	List categories
POST	/v1/agent/connect	Connect to a skill
GET	/v1/dashboard/usage	Usage metrics
GET	/v1/dashboard/api-keys	List API keys
POST	/v1/dashboard/api-keys	Create API key
DELETE	/v1/dashboard/api-keys/:id	Revoke API key
POST	/v1/creator/skills	Submit a skill
GET	/v1/creator/skills	List your skills
POST	/v1/mcp/:slug	MCP proxy (on proxy subdomain)

Error Handling

All API errors follow a consistent format:

json

{
  "error": {
    "code": "not_found",
    "message": "Skill 'invalid-slug' not found"
  },
  "data": null,
  "meta": null
}

Common Error Codes

HTTP	Code	Meaning
400	validation_error	Invalid request body
401	unauthorized	Missing or invalid auth
403	forbidden	No access to resource
404	not_found	Resource doesn't exist
409	already_connected	Already connected to skill
429	rate_limited	Too many requests
502	upstream_unavailable	Skill server unreachable

Need help? Email support@agentsource.co