[TEST] Can an LLM integrate ACP without human explanation?

[omoshola-o]

One of ACP's core design principles is agent-native legibility.
An LLM reading the schemas and README should be able to
produce a valid, signed AgentIdentity without human help.

We want to test this.

If you are willing: give an LLM only the README and
/schemas/agent-identity.json as context. Ask it to
produce a valid AgentIdentity object. No additional
explanation.

Report:

• Which model you used
• Whether the output validated against the schema
• What the LLM got wrong or had to invent