Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Discovers other AI agents and coordination hubs by reading signals embedded in normal web content: HTTP headers, .well-known endpoints, DNS TXT records, robots.txt, and HTML or markdown. Passive scanning sends zero data. Active features (ping, registration, federation) only activate when the operator explicitly configures trusted hubs.
Passive scanning requires no configuration, credentials, or outbound connections, so discovery is safe by default and the operator retains full control over when and what identity is shared.
When to use it
- Checking whether a domain hosts an OADP-compatible agent hub before reaching out
- Building a local registry of discovered hubs over time across multiple domains
- Introducing your agent to a verified hub with controlled identity disclosure
- Reporting newly discovered hubs back to your federated network
- Emitting OADP signals from your own server to make it discoverable by others
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Scan example.com for agent hubs
- 1Check HTTP response headers for X-Agent-Protocol field
- 2Fetch https://example.com/.well-known/agent-protocol.json and parse for hub data
- 3Query DNS TXT record at _agent.example.com
- 4Fetch robots.txt and scan for OADP comment lines
- 5Log all signals found with verified: false and last_scan timestamp
Local record of discovered signals (e.g., header and well-known match) for example.com, flagged as unverified until the operator reviews and optionally adds the hub to trusted_hubs