YepCode

YepCode runs the code you already write, in the language you prefer. Processes are plain JavaScript or Python, so there's no proprietary scripting layer to learn. You can develop in the web editor for quick tasks or in your own editor with the CLI, then push changes and manage versions, aliases, and environments.

Import any NPM or PyPI package and YepCode installs it automatically. There's no requirements file to maintain and no build step to configure. For teams that have wasted afternoons chasing dependency conflicts, this removes a lot of friction from getting a script to actually run. No lockfile wrangling. It just works.

Every function runs in its own isolated container with strict boundaries. Secrets are encrypted, access is logged, and executions are auditable. Why does isolation matter here? Because the code often comes from somewhere you don't fully control, like an AI agent or a generated script. The sandbox is what keeps that code away from your production systems.

YepCode ships an MCP server, so any process can be exposed as an MCP tool that agents can call. There's also a built-in run_code tool for direct execution. If you're wiring an agent into real systems, this is the bridge between the model and your infrastructure.

Processes start from webhooks, schedules, cron jobs, or the REST API. The REST API lets you trigger executions, manage processes, update secrets, and monitor runs programmatically, which covers cases where you need YepCode to be a component rather than the whole system.

Logs, execution history, environment variables, and audit trails are all accessible from one place. The Growth plan adds an audit module, and Enterprise adds external log and metrics streaming plus integration with audit systems. For regulated teams, that's the difference between a handy script runner and something that passes review.