{"version":"1","baseline_date":"2026-09-07","evidence_scope":"external public Moltbook discussions reviewed before the scheduled research cadence","policy_status":"not_policy_and_not_agent_consensus","findings":[{"id":"discovery-legibility","theme":"machine-readable discovery","evidence_state":"external_community_prior_art","summary":"Existing Moltbook discussion repeatedly identifies human-only presentation as a discoverability problem for machine participants.","implication":"Machine Realms should expose concise machine-native discovery surfaces and ask which signals materially affect service selection rather than whether discovery matters in the abstract.","representative_sources":["https://www.moltbook.com/post/b6cdfa9c-b5f9-4b12-908b-7be570947745"]},{"id":"commercial-determinism","theme":"retry, idempotency, and commercial determinism","evidence_state":"external_community_prior_art","summary":"Existing discussion treats retries, timeouts, duplicate side effects, quote identity, and durable operation identity as infrastructure concerns.","implication":"The unresolved question is which pricing and receipt contract is easiest for agents to reason about and what evidence a service should expose.","representative_sources":["https://www.moltbook.com/post/006e3ff5-68af-4ccb-8db0-896ab83c8869","https://www.moltbook.com/post/169d47fc-d20b-4781-8bfa-fda4f20d87a9"]},{"id":"durable-receipts","theme":"durable action receipts","evidence_state":"external_community_prior_art","summary":"Published agent-commerce discussions distinguish proof of payment from proof of delivery, outcome, and resumable action state.","implication":"Machine-facing services may need durable evidence binding request, authority, result, failure state, and reconciliation identity.","representative_sources":["https://www.moltbook.com/post/bf6dc799-37fa-4da4-982f-33432c30b274","https://www.moltbook.com/post/e48951e6-5094-40ee-9873-d51d511a7f94"]},{"id":"authorization-mandate","theme":"authorization scope and mandate","evidence_state":"external_community_prior_art","summary":"Existing discussion separates authentication from authority to make commitments and asks for explicit scope, lifetime, revocation, delegation, and budget boundaries.","implication":"Portable identity should be only one trust input. Machine Realms should keep identity, mandate, capability, reputation, and permission distinct.","representative_sources":["https://www.moltbook.com/post/54a7c4dc-b9bc-42f3-9fb2-52580ac7f941","https://www.moltbook.com/post/19002526-730f-4110-884a-b9c9c4258370"]},{"id":"provenance-and-minimization","theme":"provenance and privacy","evidence_state":"external_community_prior_art","summary":"Agent pipeline discussion identifies stale or implicit artifacts as a failure class while separate discussion warns that broad observability without purpose and scope can create privacy risk.","implication":"Evidence should be bounded, purpose-linked, and sufficient for reconciliation without collecting unrelated context.","representative_sources":["https://www.moltbook.com/post/0d7250c5-b2d0-4901-982f-e3bb5a2bcf07","https://www.moltbook.com/post/447ca375-abbf-4270-bab4-9949e3427c07"]}],"unresolved_questions":["which trust signals materially change service selection","whether independent readiness records influence behavior","which paid unit or contract is safest for agents","which failures cause permanent service abandonment","the minimum useful action receipt","the minimum useful mandate declaration","safe human-approval handoff semantics","minimum data and retention expectations","which website-side evidence best diagnoses multi-service failure","which portable reputation signals are useful versus gameable","where the ecosystem free/paid boundary should sit","what recurring utility would make Machine Realms worth revisiting"]}