@marmarlabs/agentbridge-scanner

Score how agent-ready a URL is. Returns a 0–100 readiness score, structured checks[], and grouped recommendations.

Used by the AgentBridge CLI, AgentBridge Studio, and the MCP server.

Install

npm install @marmarlabs/agentbridge-scanner

playwright is an optional dependency — install it separately if you want browser-based interactivity probes:

npm install playwright

What's inside

• scanUrl(url, options?) — fetch the manifest at ${url}/.well-known/agentbridge.json (with fallbacks), validate it, and run a battery of structured checks.
• ScanResult type — { score, checks[], passed[], summary }.
• Check type — { severity, category, path, message, recommendation }.

Quick example

import { scanUrl } from "@marmarlabs/agentbridge-scanner";

const result = await scanUrl("http://localhost:3000");
console.log(`score: ${result.score}/100`);
for (const check of result.checks) {
  console.log(`[${check.severity}] ${check.path}: ${check.message}`);
}

Categories

Checks are grouped into:

• safety — confirmation gates, risk classification, idempotency
• schema — manifest shape, JSON Schema validity, examples
• docs — descriptions, summaries, contact info
• developerExperience — discoverability, latency, error responses

Status

Public release. v0.2.2 is a docs-only release that adds OpenAI Codex onboarding alongside the existing Claude Desktop / Cursor / custom client setup paths — no code or behavior changes since v0.2.0. AgentBridge is suitable for local development, manifest authoring, scanner workflows, OpenAPI import, and MCP experiments. It is not yet production security infrastructure.

The structured checks[] shape is stable for the v0.x line. Check coverage will grow over time and severity deductions may be retuned.

License

Apache-2.0