AI Agents

AI agents can connect to SmoothDeploy through the Model Context Protocol (MCP) and use the same actions available in the UI. Agents act on behalf of your account and are bound by the same role permissions.

What agents can do

Build and edit tests

Agents can create tests and directories, add, update, reorder, group, and delete steps, import and export tests, generate Playwright code from a test, and manage shared steps.

Run tests

Agents can start remote runs for one test, a directory, a tag set, or an ad-hoc list of tests. They can rerun, cancel, and read results — including pass/fail stats, artifacts, screenshots, and the HTML report.

Manage schedules

Agents can list, create, update, enable, disable, trigger, and delete schedules.

Read and manage secrets

Agents can list secrets and read their metadata (name, description, version). Secret values are never returned. Agents can also create, update, and delete secrets.

Manage webhooks

Agents can list, read, update, and delete webhooks, and inspect delivery history.

Manage branches and projects

Agents can create, rename, and delete branches, and create, update, and delete projects.

Manage tags and config presets

Agents can create, update, and delete tags and config presets, including setting and unsetting the organization default preset.

Read the audit log and email inboxes

Agents can list audit log entries and read messages from email inboxes.

Read and manage API keys

Agents can list API keys, update their IP allowlist, and delete (revoke) them.

What agents cannot do

The following are intentionally not available to agents because they return secrets on creation, deliver credentials to a person, or modify organization membership:

• Create API keys
• Create or rotate webhook signing secrets
• Invite, remove, or change the role of members
• Cancel or resend invitations
• Update organization or profile settings

See Tools Reference for the complete list of available tools.