GitHub Copilot assisted permissions now generally available
Assisted permissions are now generally available in GitHub Copilot CLI, the GitHub Copilot app and VS Code. A language model reviews each permission request, auto-approving those it judges clearly safe while sending the rest for your approval. Enterprises can control the mode through policy.
Key points
- In the CLI, run /permissions and select assisted
- In the GitHub Copilot app, open the mode picker, choose Tool permissions, then Assisted
- In VS Code, open the permissions picker and select Assisted permissions
- Requests judged clearly safe proceed automatically; all others wait for your approval
- Admins can block Allow All while keeping Assisted, or disable Assisted separately
- If the default mode is blocked by policy, new sessions start with manual approval
GitHub has made assisted permissions generally available across Copilot CLI, the GitHub Copilot app and Visual Studio Code. In this mode, a language model reviews each request Copilot makes to run a command, touch a file path or open a URL. Requests it classifies as clearly safe go through automatically, and everything else waits for your approval. In the CLI you switch to it with the /permissions command.
For organizations, administrators can block Allow All while keeping Assisted available, or turn Assisted off on its own. They can also set the default permission mode for new sessions. If that default is blocked by policy, sessions fall back to manual approval. The original changelog links to the documentation covering how requests are evaluated and when approval is still needed.
Real-world use
- A developer running a long refactor who wants routine test and build commands to stop prompting for approval
- A platform team that bans Allow All but still wants safe commands handled automatically
- A developer using Copilot CLI in the terminal who is tired of confirming every shell command
Why it matters
It gives a middle ground between clicking through every prompt and granting full autonomy. Because a model makes the auto-approval decision, check your permission mode and org policy before trusting it with sensitive work.