External Integrations • MCP & Open APIs

Connect AgentBoost
to Claude Desktop, Cursor & OpenClaw.

Mount AgentBoost agents and 138+ deterministic skills into your developer desktop tools, headless microservices, or offline mobile devices with sub-5ms latency.

01

Claude Desktop & Cursor (MCP Server)

RECOMMENDED

AgentBoost natively implements Model Context Protocol (MCP). Paste this snippet into your Claude Desktop configuration file to mount skills directly.

claude_desktop_config.json
{
  "mcpServers": {
    "agentboost": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-fetch",
        "https://agentboost-seven.vercel.app/api/mcp"
      ]
    }
  }
}
02

Python Client Library

Execute deterministic skills directly from Python scripts, FastAPI backends, or LangChain pipelines.

mount_skill.py
import agentboost as ab

# 1. Connect from OpenClaw, LangChain, or Python client
client = ab.Client(api_key="YOUR_AGENTBOOST_KEY")

# 2. Mount institutional skill from 138+ registry
tool = client.skills.get('trade_lc_auditor')

# 3. Execute with sub-5ms deterministic precision
response = tool.execute(
    lc_number="LC-2026-DXB-9841",
    invoice_total=322000,
    tolerance_percent=5
)
print("Compliance Verdict:", response.data.compliance_verdict)
03

Google AI Edge Gallery (Offline Mobile)

Export and run skills 100% offline on mobile devices via Google AI Edge Gallery and Gemma 4.

  1. Open Google AI Edge Gallery on your mobile device.
  2. Navigate to Manage Skills → Import from URL.
  3. Paste the skill URL formatted as:
    https://agentboost-seven.vercel.app/api/edge-gallery/skill-md?skillId=trade_lc_auditor
  4. Execute offline with zero data latency or privacy leakage.
04

REST API / Webhooks

cURL Request
curl -X POST https://agentboost-seven.vercel.app/api/v1/execute \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "skill": "trade_lc_auditor",
    "data": {
      "lc_number": "LC-2026-DXB-9841",
      "invoice_total": 322000,
      "tolerance_percent": 5
    }
  }'