Management — API Key Management

Navigate to: Dashboard → API Key Management

What it shows: All LLM provider API keys (OpenAI, Anthropic, Google, Azure, etc.) stored in AgentWatch. Keys are encrypted at rest; only the last 4 characters are visible.

Important: These are the upstream provider keys AgentWatch uses to call LLMs — not the employee API keys used to authenticate requests to AgentWatch.

Step-by-step: Add a provider API key

1. Click API Key Management in the sidebar.
2. Click Add Key.
3. Select the Provider (e.g., OpenAI).
4. Enter a Label (e.g., prod-openai-key-1).
5. Paste the API Key.
6. Optionally set a Rate Limit (requests/minute) and Max Concurrency.
7. Click Save.
8. Click Test on the new key row — AgentWatch sends a minimal probe request to confirm the key is valid.

Step-by-step: Disable a key without deleting it

1. Find the key in the list.
2. Toggle the Enabled switch off.
3. AgentWatch stops routing requests through this key immediately. Other enabled keys for the same provider continue working.

Step-by-step: Bulk operations

1. Check the boxes on multiple key rows.
2. Click Bulk Actions → Disable Selected or Delete Selected.
3. Confirm the action.

Key status indicators:

Indicator	Meaning
Green dot	Enabled and passing health checks
Yellow dot	Enabled but last health check returned a non-200 response
Red dot	Disabled or failed authentication
Grey dot	Never tested

How to test:

1. Add a key with a deliberately wrong API key string.
2. Click Test — the row should show a red status with an error message.
3. Replace with the correct key value (click Edit) and re-test — status goes green.

API endpoints:

Method	Path	Purpose
GET	/api/providers/:id/keys	List keys for a provider
POST	/api/providers/:id/keys	Add key
PUT	/api/keys/:id	Update label, rate limit
DELETE	/api/keys/:id	Remove key
PUT	/api/keys/:id/toggle	Enable/disable
POST	/api/keys/:id/test	Test connectivity

---