Register the agent
Create its identity and associate the appropriate keys or credentials.
Loading...
Platform
Register software agents, define their permissions, and connect their actions to a known identity.
Identity & AI agents
An agent should have an identity you can distinguish from a human or another service. That identity provides a basis for permissions, delegation, and reviewing its recorded actions.
An identity does not make an AI model infallible. Permissions must be enforced by the tools and workflows the agent uses.
Platform
Create its identity and associate the appropriate keys or credentials.
Define the scopes and policies your integrated tools must enforce.
Inspect activity and revoke access when the agent should stop.
Details and limitations
Give each agent a name, purpose and accountable owner in the appropriate Console workspace. Separate the credential your software uses to call an API from the identity of the agent performing an action. Record the intended scope of delegated authority rather than treating possession of a key as permission for every task.
The identity provides a reference that receipts and review tools can connect to an agent. Your integration can then associate recorded actions with that source and inspect its status. A displayed trust score needs its underlying factors and context; it is not a substitute for an assessment.
Identity registration does not launch a process or ensure that an external tool enforces a policy. Your application and connected systems must apply their access checks. Use the documented identity and delegation contracts for the operations you actually enable.
Register a sandbox agent, create and verify a receipt, then test a suspended identity and an action outside the allowed scope. Confirm that your integration handles denial explicitly. Keep production keys separate from sandbox configuration and review access when an agent's purpose changes.
IMPLEMENTATION AND EVIDENCE
Use a server-side workspace credential to inspect agent identities and their recorded status. An identity record does not independently enforce permissions in your other tools.
Run this from your terminal or backend. Keep access tokens in your environment; never place them in public browser code. Examples use production records. Sandbox uses its own endpoint and authorized environment.
curl --fail-with-body "https://api.truthlocks.com/v1/agents" \
-H "Authorization: Bearer $TRUTHLOCKS_ACCESS_TOKEN"A JSON response with agents and total. An empty list is valid for a new workspace.
On 18 October 2026, this endpoint returned HTTP 200 for an authenticated QA workspace. This is a scoped functional check, not a performance benchmark or independent audit. Inspect the dated check record.
Business production access follows approved onboarding. Starting an account does not require purchasing a plan.