web3-agents-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| CACHE_DIR | No | Directory for the local sqlite cache of fetched registration files. | ~/.cache/web3-agents-mcp |
| LOG_LEVEL | No | One of error, warn, info, debug; controls stderr log verbosity. | info |
| INDEX_BACKEND | No | Selects the search_agents backend. Only null is implemented. | null |
| IPFS_GATEWAYS | No | Comma-separated list of IPFS HTTP gateways to try. | https://ipfs.io,https://cloudflare-ipfs.com,https://gateway.pinata.cloud |
| DEFAULT_CHAIN_ID | No | Chain id used by any tool call that omits chain. | 8453 |
| RPC_URL_<chainId> | No | Overrides/prepends the RPC endpoint for a specific chain id, e.g. RPC_URL_8453. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| pingA | Liveness check; returns pong and the server version. |
| resolve_agentA | Resolves an ERC-8004 agent by agentId or ownerAddress (exactly one selector), returning identity fields plus best-effort endpoints/capabilities parsed from its registration file. |
| get_registration_fileA | Fetches and verifies an agent's registration file (via its tokenUri): ipfs:// CIDs are verified, data: URIs are inherently verified, https:// is unverifiable in v1 (no on-chain hash commitment). |
| get_reputationA | Reads an agent's Reputation Registry feedback summary (and optionally the raw feedback entries). Always returns honesty caveats — feedback is self-reported by clients and is a weak signal, especially for low feedback counts. |
| get_validationsB | Reads an agent's Validation Registry entries (validator, best-effort method classification, response, timestamp). An agent with zero validations is a normal, successful result — not an error. |
| assess_trustA | Factual trust report for an ERC-8004 agent: runs identity, registration file, reputation, and validation lookups in parallel with graceful partial failure, and returns the raw sections plus deterministic honesty caveats and a short factual natural-language summary. No numeric scoring. |
| search_agentsA | Searches for ERC-8004 agents by name/capability/description. MVP stub: no local index backend ships yet, so this always returns INDEX_UNAVAILABLE (input validation still runs first). |
| list_chainsA | Lists the chains this server is configured for. Every other tool's |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 8 tools
Each tool targets a distinct resource or action: liveness, chain config, agent resolution, registration file, reputation, validations, composite trust, and search. assess_trust aggregates the getters but does not overlap with them, so an agent can clearly select the right tool.
All tools use snake_case and nearly all follow a verb_noun pattern (resolve_agent, list_chains, get_registration_file, get_reputation, get_validations, assess_trust, search_agents). The only exception is ping, which is a standard health-check verb and does not break the overall consistency.
8 tools is well-scoped for an ERC-8004 agent trust server. The set covers health, configuration, identity resolution, data source retrieval, a composite trust report, and search without redundancy or bloat.
The tool surface covers identity, registration file verification, reputation, validations, and a composite trust assessment. The search_agents tool is only an MVP stub that always returns INDEX_UNAVAILABLE, which limits discovery but does not block core trust workflows for known agents.