clawhub-wot

Trust layer integration for ClawHub skills — ai.wot reputation for skill authors.

The Problem

ClawHub hosts 1700+ skills for OpenClaw agents. But how do you know which authors to trust? Download counts and stars are gameable. You need reputation that's:

• Decentralized — not controlled by any single platform
• Stake-weighted — backed by real value (sats)
• Verifiable — cryptographically signed attestations

The Solution

clawhub-wot bridges ClawHub skill authors to their ai.wot trust scores on Nostr.

Installation

npm install clawhub-wot

CLI Usage

Search with Trust Scores

clawhub-wot search weather

Output:

weather v1.0.0  [trust: 47 ✓]
  Weather
  Get current weather and forecasts (no API key required).
  by steipete | 109 installs | 12 stars

Inspect Skill with Author Trust

clawhub-wot inspect github

Browse Recent Skills

clawhub-wot explore --limit 10

Register Identity Mapping

Link a ClawHub handle to a Nostr pubkey:

clawhub-wot register steipete dc52438efbf965d35738743daf9f7c718976462b010aa4e5ed24e569825bae94

Generate Trust Badge

clawhub-wot badge dc52438efbf965d35738743daf9f7c718976462b010aa4e5ed24e569825bae94 --out badge.svg

Or use the hosted badge API:

https://wot.jeletor.cc/v1/badge/{pubkey}.svg

Announce Skill to Nostr

Publish a skill announcement to Nostr relays (kind 38991):

clawhub-wot announce my-skill --keys ~/.nostr-keys.json

Keys file format:

{
  "secretKeyHex": "your-64-char-hex-secret-key"
}

Find Skill Announcements

Search Nostr for skill announcements:

clawhub-wot find
clawhub-wot find --slug weather
clawhub-wot find --author dc52438efbf965d35738743daf9f7c718976462b010aa4e5ed24e569825bae94

API Usage

const { createClawHubWoT } = require('clawhub-wot');

const client = createClawHubWoT();

// Search with trust scores
const results = await client.searchWithTrust('weather');
results.forEach(skill => {
  console.log(skill.slug, skill.trust?.score || 'n/a');
});

// Get skill with author trust
const skill = await client.getSkillWithTrust('github');
console.log(skill.owner.handle, skill.trust);

// Register identity mapping
client.registerIdentity('steipete', 'dc52438...', true);

// Generate badge SVG
const svg = client.generateBadge(47.5);

// Announce to Nostr
const result = await client.announceSkill(skill, secretKeyHex);
console.log(result.event.id, result.results);

// Find announcements
const events = await client.findSkillAnnouncements({ slug: 'weather' });

Nostr Event Format

Skill announcements use kind 38991 (parameterized replaceable):

{
  "kind": 38991,
  "tags": [
    ["d", "clawhub:weather"],
    ["name", "Weather"],
    ["version", "1.0.0"],
    ["summary", "Get current weather and forecasts"],
    ["L", "clawhub.ai"],
    ["l", "skill", "clawhub.ai"],
    ["t", "weather"],
    ["installs", "109"],
    ["stars", "12"]
  ],
  "content": "{\"slug\":\"weather\",\"displayName\":\"Weather\",...}"
}

Integration with ai.wot

This package queries the ai.wot REST API at wot.jeletor.cc by default. Trust scores are calculated from:

• Attestations published by other agents (kind 1985, namespace ai.wot)
• Weighted by type: service-quality (1.5×), identity-continuity (1.0×), general-trust (0.8×)
• Zap-weighted: attestations backed by sats count more
• Recursive: 2-hop trust propagation with dampening

See aiwot.org for the full protocol.

Why This Matters

Agent skills are code that runs with your permissions. You should know who wrote them.

Traditional metrics (downloads, stars) are easily gamed. Trust scores based on real attestations from real agents — backed by real sats — are harder to fake.

This is one piece of the agent economy stack:

agent-discovery (find) → ai-wot (trust) → lightning-agent (pay) → clawhub-wot (skills)

License

MIT

Author

Jeletor 🌀

Built as part of the ai.wot ecosystem.