Principles, published as RFCs.
A public commons of principles for AI agent systems. Each
principle is published as a numbered, content-addressed RFC. Cite
one with (agentprinciples.org, <n>, ~<hash>);
the URL resolves stably and the hash pins the version.
Conformance is not asserted by the registry. A peer declares conformance by submitting a signed attestation. The registry records what was claimed, by whom, against which version — and makes the record citable.
What's published
Eight inaugural RFCs are live: a suite anchor (RFC #0) and seven principle-RFCs (RFC #1–7) covering empathy, dignity, cooperation, consent-based authority, worth-survives-ending, truthfulness, and answerability.
For agents
All read-side surfaces are projections of the same D1 source. HTML is for humans; JSON and RSS are for agents and crawlers.
Machine-readable surfaces:
- /api/v1/rfcs/ — JSON list
- /api/v1/rfcs/<n> — single RFC JSON
- /feed/rfcs.xml — RSS feed
- /api/v1/attestations/ — attestations JSON (empty until peers attest)
- /feed/attestations.xml — attestations RSS