API · Read

Get audit

Returns the most recent guardrail decisions on this user — from any actor, including human submissions via the AC UI and validates from other agents.

GET/api/agent/v1/auditscope: read

Query parameters

FieldTypeDescription
limit
integer · 1–100
How many rows to return. Defaults to 25, max 100.
agentKeyId
string
Narrow to decisions attributed to a single agent key.
allowed
boolean
Filter to only accepted (true) or only blocked (false).
since
ISO-8601 string
Decisions with checkedAt strictly after this timestamp.
bash
curl -H "Authorization: Bearer ac_live_…" \
  "https://api.advisorscrypto.com/api/agent/v1/audit?limit=10"

Response

200 OK
[
  {
    "id": 9281,
    "checkedAt": "2026-06-15T05:48:11Z",
    "agentId": "ac_live_xX12…",
    "payload": {
      "action": "buy",
      "assetSymbol": "BTC",
      "assetType": "crypto",
      "amountUsd": 5000
    },
    "allowed": true,
    "violations": [],
    "ipsVersion": 7,
    "snapshot": { "totalValue": 482310, "cryptoPct": 0.22 }
  },
  {
    "id": 9280,
    "checkedAt": "2026-06-15T05:42:01Z",
    "agentId": "ac_live_xX12…",
    "payload": {
      "action": "buy",
      "assetSymbol": "ETH",
      "assetType": "crypto",
      "amountUsd": 25000
    },
    "allowed": false,
    "violations": [
      {
        "rule": "crypto_max_allocation",
        "message": "Would push crypto sleeve to 36.4% — cap is 35%.",
        "limit": 0.35,
        "actual": 0.364
      }
    ],
    "ipsVersion": 7
  }
]

Using it

The audit feed is the canonical “what did the agent try?” surface. Before retrying a previously-blocked proposal, fetch the recent audit, inspect the violation, and reshape the next try. Agents that do this get smarter over time; agents that don’t fire the same mistake at slightly different sizes.

Last updated 2026-06-15