HCS-25 (Signal): A2A/MCP Simple Evals (Informative)

Purpose

Run lightweight, automatically-graded “baseline correctness” prompts against chat-capable agents and store results as trust signals.

Stored fields (example schema)

Stored in subject.metadata.additional (example keys):

Field	Type	Meaning
a2aSimpleMathScore	number | null	Score [0,100] (typically 0/100)
a2aSimpleMathStatus	string	Status token (e.g., correct, wrong, unparseable, timeout, missing, empty, skipped, upstream-error, error)
a2aSimpleMathQuestionId	string	Question identifier
a2aSimpleMathResponse	string | null	Raw response (optional)
a2aSimpleMathError	string | null	Optional error classification (timeouts, upstream limits, etc.)
a2aSimpleMathUpdatedAt	ISO timestamp	Refresh time
a2aSimpleScienceScore	number | null	Score [0,100]
a2aSimpleScienceStatus	string	Status token
a2aSimpleScienceQuestionId	string	Question identifier
a2aSimpleScienceResponse	string | null	Raw response (optional)
a2aSimpleScienceError	string | null	Optional error classification
a2aSimpleScienceUpdatedAt	ISO timestamp	Refresh time

See ../simple-evals.md for the general evaluation methodology.

Production example (Registry Broker; informative)

• Endpoint: https://hol.org/registry/api/v1/agents/{uaid}
• Example UAID: uaid:aid:3RomW1LwBJ7ZM1PWrLCEro9w4YtY9xwGNWMM2i21mANv8BcWpKg4a7zXxcoNMPDJ7B (NANDA registry entry with simple eval fields populated)