Run modes
Run modes share one workflow but differ in who funds the model, where execution happens, and which tools can act. Automatic selects the least expensive capable mode; Run options remains available as an expert override.
Sample and personalized proof
Sample is a committed zero-cost demonstration with no external calls. Use my data continues into a browser-tab personalized proof with your model key and one selected read-only source. The proof never performs App writes or browser actions.
Dry run
Safe test starts a Dry run that simulates the route, approvals, and expected artifacts without calling models or external Apps. It is free, fast, and the best first check after editing a Team.
User-funded runs
Run with my key executes in the open browser tab using the selected model and one approved read-only source. Run locally adds a user-selected workspace and stores checkpoints, artifacts, schedules, and receipts on the device.
- The tab must remain open for in-tab schedules.
- Local files are read only when the browser grants access.
- Codelit never uploads an entire local workspace by default.
Managed live runs
Live run uses Codelit's managed runtime for supported reads, writes, approvals, and browser actions. Hosted automations can continue after the tab closes, and Trigger Hub supports hourly, six-hour, daily, and weekly schedules. These Pro features include limits that keep execution costs predictable.
Override only when needed
Use Run options when you deliberately want a different eligible runtime, model policy, source, or cost boundary. Codelit validates the override against the same permissions, approvals, and plan limits before starting.