Getting Claude Code (CC) to work autonomously without permissions issues
The user is trying to set up a development loop with Claude Code (CC) to automate JIRA issue creation and execution, but is facing issues with permissions and tools being repeatedly blocked or asked for approval.
Why it matters
Automating JIRA issue management can significantly improve productivity and efficiency, but the permission issues the user is facing are preventing them from fully realizing the benefits of this setup.
Key Points
- 1The user has set up a development loop with multiple subagents to automate JIRA issue management
- 2CC is able to execute commands like 'gh pr create' initially, but then starts asking for permissions again
- 3The user has tried using the Sandbox feature, but it caused other issues like blocking GitHub access
Details
The user is trying to set up a fully autonomous workflow using Claude Code (CC) to plan, create, and execute JIRA issues without any manual intervention. They have managed to get the setup working, where CC can handle the entire process end-to-end. However, they are facing issues with permissions, where CC is able to execute certain commands initially, but then starts asking for approval again, even for commands that were previously approved. The user has tried using the Sandbox feature to isolate CC, but that caused other problems like blocking access to GitHub. They are looking for a solution to run CC autonomously without constantly running into permission-related roadblocks.
No comments yet
Be the first to comment