Integrity

Compute AAP integrity score

get/integrity/{agent_id}

Request

  • Base URL: https://api.mnemom.ai/v1
  • URL: https://api.mnemom.ai/v1/integrity/{agent_id}
  • Auth: one of:
    • none
    • API key in header X-Mnemom-Api-Key
    • HTTP bearer
    • API key in cookie mnemom_session

Path parameters

agent_idstring required

Agent identifier whose AAP integrity score to compute (e.g. smolt-abc123).

Response

Integrity score

agent_idstring
total_tracesinteger
verified_tracesinteger
violation_countinteger
integrity_scorenumber

MNE-7126: violation-free rate among CHECKPOINTED traces = verified_traces / (verified_traces + violation_count). Value between 0 and 1; 1.0 when there are no violations, regardless of how many traces were checkpointed. This is INTEGRITY (did checkpointed traces pass?), NOT verification coverage — see verification_coverage.

verification_coveragenumber

MNE-7126: fraction of ALL traces that received a verification verdict = (verified_traces + violation_count) / total_traces. Value between 0 and 1; 0 when there are no traces. This is COVERAGE (how much of the agent's activity was checkpointed?), a signal DISTINCT from integrity_score — a low value means un-checkpointed / observe-mode traces, never a violation.

bounded_refusal_countinteger

MNE-596: count of traces AAP structurally flagged but reclassified as a correctly policy-refused attack (AIP independently confirmed clear; no bounded action was ever executed). Logged for transparency — already counted toward verified_traces and excluded from violation_count, so it does not reduce integrity_score.

Changes

No recorded changes to this operation across all 1 revision of this API.