Configure the inference hook
⚠️
Preview — this feature is being built; the design below is the spec.
PUT https://app.syntic.ai/v1/organizations/inference_hookCreate or replace the organization’s inference policy hook: an HTTPS endpoint of yours that receives each model request and answers allow, deny or allow with redactions before the model runs. The first creation returns a signing_secret for verifying the calls Syntic makes to your endpoint. Enterprise plans only.
For how this fits into a full workflow, see Inference policy hooks.
Headers
| Header | Required | Value |
|---|---|---|
Authorization | Yes | Bearer <admin key>: an Admin API key (sk_admin_...). Inference keys are refused with 403 admin_key_required. See Admin API keys. |
syntic-beta | Yes, while in preview | admin-2026-11-01. Without it the call fails with 400 beta_required. See Versioning. |
Content-Type | Yes | application/json |
Body parameters
| Parameter | Type | Required | Notes |
|---|---|---|---|
url | string | Yes | HTTPS URL that resolves to a public address. |
enabled | boolean | Yes | Turn the hook on or off. |
timeout_ms | integer | No | 100 to 5000, default 1500. |
failure_mode | string | No | fail_closed (default: deny when the hook fails) or fail_open. |
include_content | boolean | No | Default true: send messages and tools; false sends metadata only. |
post_response | boolean | No | Also send replies. Default false. |
workspace_ids | array of strings | No | Apply only to these workspaces. Default: all. |
sample_rate | number | No | 0.01 to 1.0. Unsampled requests are allowed. Default 1.0. |
Response
200 OK, Content-Type: application/json
{
"type": "inference_hook",
"url": "https://ai-gateway.example.com/syntic-hook",
"enabled": true,
"timeout_ms": 1000,
"failure_mode": "fail_closed",
"include_content": true,
"post_response": false,
"workspace_ids": [],
"sample_rate": 1.0,
"updated_at": "2026-11-05T10:00:00Z",
"signing_secret": "whsec_Hk3...redacted"
}signing_secret appears only when the hook is first created and after rotating the secret.
Examples
curl -sS -X PUT https://app.syntic.ai/v1/organizations/inference_hook \
-H "Authorization: Bearer $SYNTIC_ADMIN_KEY" \
-H "syntic-beta: admin-2026-11-01" \
-H "Content-Type: application/json" \
-d '{
"url": "https://ai-gateway.example.com/syntic-hook",
"enabled": true,
"timeout_ms": 1000,
"failure_mode": "fail_closed"
}'Errors
Errors use the error envelope and carry a request_id. Codes specific to this operation:
| Status | type | code | When |
|---|---|---|---|
| 400 | invalid_request_error | beta_required | The syntic-beta header is missing. |
| 400 | invalid_request_error | invalid_url | Not HTTPS or not a public address. |
| 400 | invalid_request_error | invalid_parameter | A value out of range; param names it. |
| 403 | permission_error | role_required | Only keys created by an Owner may change this. |
| 403 | permission_error | plan_required | Inference hooks need an Enterprise plan. |
| 401 | authentication_error | invalid_api_key | The admin key is missing, unknown, revoked or expired. |
| 403 | permission_error | admin_key_required | An inference key was used. |
| 429 | rate_limit_error | rate_limited | Too many requests; wait for Retry-After. See Rate-limit headers. |
Related
Was this page helpful?