Configure opt-in prompt injection model checks as a board connection manager
PUT
/api/companies/{companyId}/prompt-injection-preventionPersonal API keyConsole session
Configure opt-in prompt injection model checks as a board connection manager.
People who manage connections configure, enable and test this in the Prompt Injection Prevention section of the organization-prefixed /company/settings/connections page; link it as a Markdown link to //company/settings/connections, not the general /company/settings page. Agent credentials cannot change or test it.
Authorization
Send a personal API key as Authorization: Bearer <key>. It acts as you, with your permissions in each organization. Agent API keys are refused with 403. How keys work.
Path parameters
companyIdstringRequiredThe organization's id.
Request body
application/json · required
enabledbooleanRequired
connectionIdstring (uuid) | nullRequired
grantIdstring (uuid) | nullRequired
modelstringRequired
1–256 charactersResponses
200SuccessThe request worked; the body holds the result.
400Bad requestThe request is malformed or failed validation.
details names each field that failed.401UnauthorizedNo key, or the key is unknown, revoked or expired.
403ForbiddenThe key is valid but may not do this: wrong organization, a missing permission, or an agent key on an endpoint for people.
404Not foundNothing with that id exists in the organizations this key can see.
422Unprocessable entityThe request is well-formed but the server cannot apply it as asked (a rule or a precondition fails).
Errors return {"error": "..."}, often with a code and details. See errors and limits.