Secrets & Settings
How to manage secrets, provider keys, and runtime settings in the dashboard.
Secrets
The Secrets tab stores runtime credentials and environment variables. Values are encrypted at rest. Provider keys (Anthropic, OpenAI, etc.) are never injected directly into the container — AgentStead injects a per-agent proxy token instead and forwards requests to the real provider at the network layer. See Credential isolation for how this works.
Secret values are masked after saving. Copy anything you'll need again before closing the tab.
Adding a secret
- Open your agent → Secrets tab
- Enter the key name and value
- Save
Common keys:
ANTHROPIC_API_KEY=sk-ant-...
OPENAI_API_KEY=sk-...
GEMINI_API_KEY=...Bulk import
Paste multiple secrets in .env format and save them all at once:
ANTHROPIC_API_KEY=sk-ant-...
OPENAI_API_KEY=sk-...
MY_WEBHOOK_URL=https://hooks.example.com/...Each line becomes a separate secret.
Secret lifecycle
Rotating a key
Delete the old entry
Go to Secrets, find the key, and delete it. The old value is permanently removed immediately.
Add the new value
Add a new secret with the same key name and the new value.
Restart the agent
Go to the agent's overview and click Restart. The new secret is injected on startup.
Runtime settings
The Settings tab holds structured runtime configuration — model/provider settings and platform-managed defaults.
In practice:
- Secrets = sensitive values (API keys, tokens, passwords)
- Settings = structured configuration (model selection, provider defaults)
Some settings changes take effect immediately; others need a restart. If in doubt, restart.