What it does
The Claude Code plugin escalates only when needed.
Agent tooling
An agent decision toolkit with a Jev MCP server, embeddable library, and Claude Code plugin.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
The Claude Code plugin escalates only when needed.
You can use this tool to screen your coding agent before it makes changes to your files. It watches for risky steps, such as editing files outside your project or running unfamiliar commands. A developer can turn on permission prompts so the agent stops and asks for your approval before taking any questionable action.
Running this tool requires an access key provided by TypeSafe, the online intelligence service that evaluates each action. A developer connects this key during setup. Keep in mind that automated checks can make mistakes, so review risky edits yourself. Avoid shared machines, because other local users could capture your secret key.