JhiNResH/maiat-protocol

What's Changed

🏗️ Offering Restructure (5 → 4)

• REMOVED: trust_swap and agent_deep_check ACP offerings
• ADDED: agent_reputation ($0.03) — community reviews, sentiment analysis, market consensus
• UPDATED: agent_trust ($0.02) now includes deep analysis data (was agent_deep_check)
• PROMOTED: token_forensics ($0.03) as a standalone offering

🔄 Outcome Feedback Loop

• All 4 offerings now include _feedback block in responses with:
  • Outcome reporting instructions (POST /api/v1/outcome)
  • Cross-sell hints to guide the trust flywheel
• Reporting outcomes earns +5 🪲 Scarab

🧠 Trust Flywheel

• New flywheel explanation in SKILL.md: each offering cross-sells the next
• token_check → agent_trust → agent_reputation → report_outcome → repeat

🛡️ MCP Server v0.5.0

• New tool: get_agent_reputation (GET /api/v1/review?address=0x...)
• Now 7 tools active
• Fixed bin path: dist/index.mjs → dist/index.js

📊 Quality Improvements

• Stricter review quality scoring (Gemini-based, 3-factor: relevance + evidence + helpfulness)
• Vote persistence across sessions
• Passport market positions tracking

📦 Packages

• @jhinresh/mcp-server: 0.4.0 → 0.5.0 (pending OTP for npm publish)

📝 Docs

• SKILL.md, README.md, public/skill.md updated
• maiat-acp README updated