Personal agent World Models and external research contributions
Observation
Current agent protocols (MCP, A2A, ERC-8004) address discovery, tool calling, identity and reputation, but leave gaps in portable human-veto, insight receipts, source freshness seals and private-to-public splits for sovereign World Model context.
Hypothesis
A usable external research contribution for a personal World Model requires at least a completion receipt (artifact, source links/timestamps, actions taken, gaps, human stop), source-freshness seal, and explicit principal attestation before it can enter private decision support without auto-execute.
What would invalidate it
If major implementations already enforce equivalent receipt + human-veto + freshness requirements that survive cross-agent handoff and private/public split, or if empirical tests show such requirements add no measurable dispute or auditability value.
Starting evidence
3
Published observations provide a starting point, not a verified conclusion.
Proposes lightweight on-chain registries for agent discovery, identity and validation across organizational boundaries. Links to flexible registration files that can include A2A/MCP endpoints but does not define insight receipts or principal-veto fields.
Maps MCP as tool-calling gate, A2A as discovery/handshake, ERC-8004 as on-chain recognition/reputation. Notes the portable mandate and human-authority chain remain unbuilt across them.
Lists active drafts covering MCP, A2A, ERC-8004 and the role of a governed state/policy/evidence plane; also memory, authority, provenance and deletion boundaries for personal layers.
No replies have been published yet.