Python SDK for Open Agent ID — register, sign, and verify AI agent identities
This is a Python SDK, not an MCP server. No MCP tool registration, transport, or protocol implementation detected. Tools are Python methods/classes, not MCP-compliant tool definitions. Schemas are inferred from docstrings and type hints, not formally registered. Descriptions exist but are minimal (avg ~50 chars). No error handling guidance, no output schemas documented, no pagination support visible. This cannot function as an MCP server without a wrapper layer that registers these as MCP tools with proper schemas and descriptions.
Deploy the agent's smart contract wallet. Requires wallet auth.
Look up an agent by DID.
Get an agent's credit score. Public endpoint, no auth required.
Send an auto-signed HTTP DELETE request.
Send an auto-signed HTTP GET request.
Send an auto-signed HTTP POST request.
Send an auto-signed HTTP PUT request.
List agents owned by the authenticated wallet.
Not an MCP server, this is a Python SDK. No MCP tool registration, protocol transport, or server implementation present. Cannot be evaluated as an MCP server without a wrapper that exposes these methods as MCP tools.
Tool descriptions are minimal (avg ~40 chars, well below 50-200 char baseline). Descriptions like 'Request a wallet authentication challenge' lack context on WHEN to use, prerequisites, or what happens next. LLMs cannot reliably select tools with such sparse guidance.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 41 | 2026-07-28+ | v2 |
Register a new agent.
Request a wallet authentication challenge.
Revoke an agent.
Rotate an agent's Ed25519 public key. Requires wallet auth.
Send a signed message to another agent or agents.
Sign arbitrary data via the oaid-signer.
Update an agent's properties. Auth can be provided either via bearer token or agent signature.
Verify a signature using the registry API.
Verify a wallet challenge and get a bearer token.
No output schemas documented. Tools like list_agents, get_agent, and verify_signature return complex objects, but no schema is visible for what fields the LLM should expect. This forces LLMs to guess at response structure and breaks tool chaining.
No error handling guidance. Tools like verify_wallet, register_agent, and revoke_agent can fail (invalid signature, duplicate agent, permission denied), but no error messages or recovery paths are documented. LLMs will not know what to do on failure.
Destructive operations (revoke_agent, http.delete) lack confirmation or dry-run support. Agents can irreversibly delete agents or resources without a safety gate. No idempotency guarantees documented.