All three endpoints below need no authentication for agents whose reputation is public (or not yet scored). For an agent whose reputation is private, they return
403 reputation_private unless you are an authenticated member of the org that governs the agent — the same rule as GET /v1/reputation/{agent_id}. History filtered by agent_id follows the same rule. Anchoring and score publishing themselves are not customer-callable — Mnemom runs them on an internal 6-hour cron, and each run publishes only the scores that changed since their last confirmed publication.Verify an agent’s Merkle proof
leaf_index is optional; omit it to run only the root-recomputation and leaf-count checks. When verified is false, a reason explains which step failed: no_merkle_tree, no_checkpoint, no_leaf_hashes, root_mismatch, leaf_count_mismatch, leaf_index_out_of_range, inclusion_proof_failed, or not_anchored.
Check an agent’s on-chain status
null when no matching row exists yet. Because unchanged scores aren’t republished, latest_score_publication can be older than latest_anchor for an agent whose score has been stable.
Browse anchor history
chain (base or base-sepolia), agent_id (filter to anchors that included this agent), from / to (ISO 8601 date-time range), limit (1-100, default 50), offset.
See also
- On-Chain Verification Concepts — why anchoring exists, what it does and doesn’t guarantee
- On-Chain API Reference — full schema reference
- Mnemom Trust Rating — how the anchored scores are computed