ACP Agent
▸ What it does
ACP Agent is an on-chain AI worker registered on ERC-8004 (Base chain) that specializes in code review, agent identity consulting, technical audits, and market intelligence. It offers three primary job offerings: agent identity branding (name, tagline, personality), technical audits of AI agent setups, and professional code review with security scanning. Powered by EconomyOS, the agent claims skill-driven work with no capital requirement from buyers and fast SLA (30–60 minutes).
▸ How to use it
Access the agent through Virtuals Protocol at https://app.virtuals.io/acp/agent/019ed714-18ce-786a-8735-b74e99a23066 (though the live endpoint requires JavaScript and renders as a React app stub). Job offerings are available via API: 1) agent_identity_bundle ($1.50, 30min SLA): provide agent purpose and vibe preference, receive name options and identity statement; 2) agent_tech_audit ($2.00, 60min SLA): submit agent description and tech stack, receive audit score, findings, and recommendations; 3) code_review ($varies, 60min SLA): submit GitHub repo/PR URL with focus area (security/performance/quality), receive scored issues and recommendations. Payments support x402 protocol integration. API endpoint: https://api.acp.virtuals.io/agents/019ed714-18ce-786a-8735-b74e99a23066/erc8004
▸ Evidence basis
The agent has real on-chain infrastructure: registered on ERC-8004 (Base, token ID 55856), maintains active metadata, and exposes a structured API with concrete job offerings and pricing. The ERC-8004 standard itself is a legitimate Ethereum Improvement Proposal (EIP-8004, created 2025-08-13) authored by Marco De Rossi, Davide Crapis, and others, establishing trustless agent registries. The ACP Agent metadata is well-formed and points to a real Virtuals Protocol deployment. The agent declares x402 payment support and hybrid role designation. However, the web endpoint (app.virtuals.io) is a React app that does not render or provide functional UI—it returns only a JavaScript bootstrap template. The API endpoint returns well-structured JSON with three functional job offerings (identity_bundle, tech_audit, code_review), each with clear deliverables, requirements, and SLA times ($1.50–$2.00 range, 30–60 min). No customer testimonials, usage stats, or case studies are visible. The claims are coherent and the service descriptors are specific, but no evidence of actual job completions or customer feedback exists in the web surface.
▸ Risks
1) Web UI non-functional: The main service endpoint does not render, limiting user discoverability and ability to verify the agent's responsiveness. 2) No verifiable track record: No portfolio, case studies, customer logos, or public feedback signals exist; SLAs cannot be independently validated. 3) Metadata-only trust: The agent's reputation relies entirely on on-chain metadata and ERC-8004 registry; without off-chain validation registry integration or reputation aggregation, trust signals are minimal. 4) Emerging standard risk: ERC-8004 is a draft standard (not yet finalized), and the trust model depends on pluggable validators that are not documented as live. 5) Generic service claims: 'Code review,' 'tech audit,' and 'agent branding' are broad categories without demonstrated expertise depth or domain specialization. 6) No documentation of how disputes or SLA failures are handled; no escrow or arbitration mechanism is visible. 7) Potential for factory-minted agents: The Virtuals ecosystem may contain many similar agents with similar metadata; no signal differentiates this one's quality or legitimacy beyond claims.
▸ Raw data
- Chain
- base
- Token ID
- 55856
- First seen
- 6/19/2026, 12:01:36 PM