Skip to content

Runs the guardrails over a submitted piece of agent text.

POST
/v1/m/security/guardrails/inspect
curl --request POST \
--url https://example.com/v1/m/security/guardrails/inspect \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data '{ "agent_ref": "example", "enforce": true, "resource_ref": "example", "session_ref": "example", "surface": "example", "text": "example" }'

Runs the guardrails over a submitted piece of agent text. It is the inline INSPECTION seam a caller (a model gateway, an orchestrator, an agent integration) uses; the module detects, produces findings with minimal evidence, and returns a DETECTIVE verdict — or a BLOCK only when inline enforcement is enabled and governed.

X-Olivares-Tenant
string format: uuid

Target tenant id; required when the principal can act in more than one tenant.

Media type application/json
object
agent_ref
Any of:

Trimmed and byte-capped.

string
enforce
Any of:

Requests blocking only when the tenant’s governed policy enables it.

boolean
resource_ref
Any of:

Trimmed and byte-capped.

string
session_ref
Any of:

Trimmed and byte-capped.

string
surface
required

After trimming, must be input, output, tool_args or tool_result.

string
text
Any of:

Inspected only in memory, never stored raw; capped at 1048576 UTF-8 bytes.

string
Example generated
{
"agent_ref": "example",
"enforce": true,
"resource_ref": "example",
"session_ref": "example",
"surface": "example",
"text": "example"
}

OK

Media type application/json
object
Example generated
{}

Bad request

Media type application/json
object
Example generated
{}

Unauthenticated

Media type application/json
object
Example generated
{}

Forbidden

Media type application/json
object
Example generated
{}

Not found

Media type application/json
object
Example generated
{}

Conflict / setup required

Media type application/json
object
Example generated
{}

Rate limited

Media type application/json
object
Example generated
{}