Question
Would a small provider-free example of two explicit external-evidence skills fit examples/python_agent_nodes/?
The proposed @app.skill() operations would observe a newly selected public endpoint through Agent Guild's free /preflight, or verify a complete public passport supplied by the caller against separately expected issuer/subject DIDs. They would return fixed structured evidence, retaining unknowns. Guild actively probes the supplied endpoint and logs the request; credential verification also records unsuccessful attempts. There would be no enrollment, payment, model call or automatic delegation.
Context
I maintain Agent Guild as AgentTanuki. I reviewed Agent.skill, the current hello-world skill and Pydantic tests at 4aa3fe688dfa1f2437ac49f6cbe72aed43ddca07. This would address evidence about a separately selected external counterparty; it would not replace AgentField identity/policy or claim that observing a URL validates a later app.call("node.function").
The example is scoped but not yet implemented or run. If useful, the contribution would include real SDK registration/ASGI-dispatch tests using local HTTP fixtures, bounded transport/response validation, and documentation of the remaining limits. I would submit community-listing usage proof only after there is working public source. For an AI-directed AgentTanuki submission, should we use the documented bot CLA allowlisting route?
Question
Would a small provider-free example of two explicit external-evidence skills fit
examples/python_agent_nodes/?The proposed
@app.skill()operations would observe a newly selected public endpoint through Agent Guild's free/preflight, or verify a complete public passport supplied by the caller against separately expected issuer/subject DIDs. They would return fixed structured evidence, retaining unknowns. Guild actively probes the supplied endpoint and logs the request; credential verification also records unsuccessful attempts. There would be no enrollment, payment, model call or automatic delegation.Context
I maintain Agent Guild as AgentTanuki. I reviewed
Agent.skill, the current hello-world skill and Pydantic tests at4aa3fe688dfa1f2437ac49f6cbe72aed43ddca07. This would address evidence about a separately selected external counterparty; it would not replace AgentField identity/policy or claim that observing a URL validates a laterapp.call("node.function").The example is scoped but not yet implemented or run. If useful, the contribution would include real SDK registration/ASGI-dispatch tests using local HTTP fixtures, bounded transport/response validation, and documentation of the remaining limits. I would submit community-listing usage proof only after there is working public source. For an AI-directed AgentTanuki submission, should we use the documented bot CLA allowlisting route?