Understand API-key security and company scope
A TimeSentry key does not merely identify software. It authenticates a company actor, including an individual, firm, or AI agent. The backend applies that actor's company scope and relationships to every request.
Use least privilege by designCreate separate keys for each application, agent, environment, and automation. Use descriptive names. Do not share a key between production and testing or between unrelated agents.
For AI agents, handler relationships, read grants, assigned tools, and bound credentials add separate authorization boundaries. Review all of them; rotating the TimeSentry key does not revoke an external credential stored for a tool.
Store and transmit securelyUse an approved secret manager or the client's protected secret field. Never commit keys to a repository, store them in ordinary notes, expose them in URLs, or send them through email or chat. Redact keys from logs and screenshots.
Audit and rotateReview creation and last-used dates. Revoke keys that are stale, unexpectedly active, or owned by departed staff. Rotate an agent key from its workspace when exposure is suspected; the previous key stops working.
Incident responseRevoke first, then investigate. Identify the key prefix, affected company, last-use window, clients or agents using it, and any created or changed records. Replace the key only in authorized systems.
Related articles: Create and revoke TimeSentry API keys, Configure an agent's handlers and permissions, Security and data privacy overview.