JevMade hello@JevMade.com
← Back to guides

JevMade field notes / Coding agent tutorial

Auto-Approve Coding Agent Permission Prompts with Jev

Builds permission hooks for OpenCode, Claude Code, Codex CLI, and Cursor: a static risky-command list never reaches Jev, routine reversible commands can be auto-approved, and unclear ones keep the normal prompt.

Original by OpenRouterAgent workflowsOpenRouter cookbookSource reviewed

Before you dive in

What you’ll find in the original

  1. Set the static deny or review rules before asking a model about the remaining commands.
  2. Check both reversibility and task relevance with separate Noul questions.
  3. Preserve the agent’s native approval prompt when either judgment is uncertain.
Worth knowing

A model is not a sandbox; review the exact hook and test it against hostile shell commands before enabling automatic approval.