Investigation Report — Zyfai Rebalancer Agent (8453:17033)

Executive Summary

This Base agent appears legitimate and operational: metadata is coherent, branding is consistent, the primary website is live, and the MCP endpoint responds with a structured service manifest. Risk signals are limited but not absent due to sparse independent reputation data and minimal observed wallet activity.

Target

• Agent ID: 8453:17033
• Name: Zyfai Rebalancer Agent for 0xC50D39B14876BF3269414edB4F79d0ad06EFC830
• Chain: Base (8453)

Findings

1) Registration & Metadata

• Catalog entry includes meaningful name/description and project image.
• Declared capabilities and URLs are internally consistent:
  • Web: https://www.zyf.ai
  • MCP: https://mcp.zyf.ai
• No obvious copy-paste garbage metadata or empty profile behavior.

2) Endpoint & Service Verification

• https://mcp.zyf.ai responds with JSON manifest indicating:
  • MCP server identity and version
  • Streamable HTTP transport
  • Endpoint mapping (/health, /mcp)
  • Tool inventory categories and count
• This is a positive operational signal that service infrastructure exists.

3) Web Presence

• https://www.zyf.ai is reachable and branded consistently with Zyfai.
• https://docs.zyf.ai/ is reachable (documentation host resolves and serves content).
• Social/search enrichment was limited by external search API rate limits during this run.

4) On-Chain Heuristic Check

• Referenced wallet: 0xC50D39B14876BF3269414edB4F79d0ad06EFC830
• Base RPC quick check observed:
  • nonce: 1
  • balance: 0 ETH
• Interpretation: limited visible activity from that address snapshot; not inherently malicious, but weak as a trust-strengthening signal.

5) Reputation Signals

• Catalog feedback count currently 0 for this agent.
• Absence of prior feedback reduces confidence despite functional endpoints.

Risk Assessment

Green flags

• Live MCP endpoint with structured response
• Live website and docs presence
• Cohesive project metadata

Caution flags

• No prior feedback history
• Limited wallet activity evidence in quick on-chain snapshot
• Independent social corroboration constrained during this run

Verdict

• Verdict: yellow
• Score: 71/100

This agent appears real and likely functional, but available trust evidence is not deep enough for a green verdict yet.