Skip to content
WebMCP Challenge · Hackathon Submission

AgentBadge — First Agent-Native Compliance Platform Powered by WebMCP

Six imperative tools, a declarative form API, and a discovery endpoint — all exposed through the browser's native document.modelContext interface. AI agents discover and use AgentBadge without leaving the page.

View Dashboard DataHub Integration
6 Imperative Tools 1 Declarative Form Discovery Endpoint

Imperative Tools

Six tools exposed via document.modelContext.registerTool()

agent-readiness-scan

Scan a website for AI agent readiness compliance. Returns a compliance score and list of checks.

readOnly untrusted
GET /api/scan?url=

badge-generate

Generate a compliance badge SVG for a given URL and score.

readOnly trusted
GET /api/badge?url=

passport-issue

Issue an AgentBadge passport NFT for a given account on Hedera.

write trusted
POST /passport/request

passport-verify

Verify an AgentBadge passport by token ID. Returns on-chain identity and metadata.

readOnly trusted
GET /api/passport/verify?tokenId=

get-compliance-score

Get the agent readiness compliance score for a given URL.

readOnly untrusted
GET /api/score?url=

search-rules

Search agent readiness rules by keyword. Returns matching rule definitions.

readOnly trusted
GET /api/rules/search?q=

Declarative API

Form-based tool exposure for agents that prefer declarative discovery.

submitScanRequest

Submit a scan request via HTML form. The agent fills the form and the browser submits it.

action /api/scan
method GET
input url (string)

Live Form

Why WebMCP?

No API keys needed. Tools are discovered natively by the browser and exposed to AI agents through the standard document.modelContext interface.

Secure by default. Trusted tools require user consent. Untrusted tools run in sandboxed contexts. The browser mediates all calls.

Zero integration. Agents discover tools automatically via the discovery endpoint. No manual configuration or MCP server setup required.

Try It Now

WebMCP is a W3C proposal — test the endpoints directly today

Test via curl

All six tool endpoints are live HTTP APIs. Test them directly:

curl -s "https://agentbadge.xyz/api/scan?url=https://example.com" | jq .

curl -s "https://agentbadge.xyz/api/score?url=https://example.com" | jq .

curl -s "https://agentbadge.xyz/api/rules/search?q=robots" | jq .

curl -s "https://agentbadge.xyz/api/badge?url=https://example.com" | head -3

Declarative Form

The form above works as a normal HTML form. Enter a URL and click Scan — no WebMCP support needed.

When browsers implement document.modelContext, the inject script will automatically register all 6 tools. The page is spec-ready.

Discovery Endpoint

Agents discover available tools via the WebMCP discovery manifest.

GET /.well-known/webmcp.json 
{
  "tools": [
    {
      "name": "agent-readiness-scan",
      "description": "Scan a website for AI agent readiness compliance.",
      "readOnly": true,
      "untrusted": true,
      "endpoint": "/api/scan"
    },
    ...
  ]
}
View live discovery manifest

Agent-Native Compliance Starts Here

AgentBadge exposes six imperative tools and one declarative form through WebMCP. AI agents discover, scan, verify, and issue passports — all without leaving the page.

View Dashboard Agent Guide Back to Home