Safe House

Resolve a human-tier review-hold (allow/deny/escalate)

The human decider for a Safe House review-hold (Wave 4). Customer/human auth (an authenticated org member, or platform staff) — the resolver is ALWAYS the authenticated principal, NEVER read from the body. A review_id outside the caller's org fleet returns 404, never 403 (no existence leak). allow/deny finalize the hold through the gateway (a NEW Ed25519-signed decision row; on allow, the decrypted original plus a single-use, content-bound allow-token are returned). escalate is non-terminal: it records the action, signals higher-tier attention, and leaves the hold pending. When the agent's card requires dual-control, a terminal ruling needs two distinct resolvers to concur — a same-identity second action returns 409, and divergent votes escalate.

post/safe-house/reviews/{review_id}/resolve

Request

  • Base URL: https://api.mnemom.ai/v1
  • URL: https://api.mnemom.ai/v1/safe-house/reviews/{review_id}/resolve
  • Auth: one of:
    • HTTP bearer
    • API key in header X-Mnemom-Api-Key

Path parameters

review_idstring required

Request body

decision'allow' | 'deny' | 'escalate' required

allow/deny finalize (single-control) or count toward dual-control concurrence; escalate is non-terminal.

reasonstring

Optional free-text rationale recorded on the approvals ledger.

Response

The resolution outcome. Terminal on a finalized allow/deny (with the allow-token + decrypted content on allow); otherwise pending (escalated, or awaiting a second dual-control approver).

review_idstring required
status'allowed' | 'denied' | 'pending' required
allow_tokenstring

Single-use, content_hash-bound, short-TTL token honored by the gateway on re-submit. Present only on a finalized allow.

contentstring

The decrypted original held content. Present only on a finalized allow.

awaiting'second_approver'

Present when dual-control is required and this is the first concurring resolver.

escalatedboolean

True when the action escalated the hold for higher-tier attention (explicit escalate, or divergent dual-control votes).

reasonstring

Changes

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