A VSCode extension that automatically approves Claude Code tool requests matching user-defined regex patterns.
When Claude Code asks for permission to run a command, Auto-Authorize checks it against your rules and instantly approves matching requests. Commands that don't match any rule are shown as normal approval prompts.
Auto-Authorize patches child_process.spawn in the VSCode extension host to intercept Claude Code's CLI subprocess communication. Permission requests (can_use_tool) are parsed from the newline-delimited JSON stdio stream, evaluated against your regex rules, and either auto-approved or passed through to the normal prompt.
- Download the latest
.vsixfrom the Releases page - In VSCode: Extensions view >
...menu > Install from VSIX... - Reload the window
Or build from source:
git clone https://github.com/c0d1f1ed/AutoAuthorize.git
cd AutoAuthorize
npm install
npm run compile
npm run packageThen install the generated auto-authorize-0.1.0.vsix.
- Click the shield icon in the activity bar to open the Auto-Authorize panel
- Go to the Rules tab and click + Add Rule
- Choose an action: Allow (auto-approve), Ask (prompt user), or Veto (silently deny)
- Select a tool type (Bash, Read, Write, Edit, WebFetch, WebSearch, or Any)
- Choose a scope: Workspace (this project only) or Global (all workspaces)
- Enter a regex pattern (e.g.,
^(grep|ls|cat|head|tail|wc)\bfor read-only commands) - Add a description and click Save
Rules are evaluated in three tiers: veto first, then ask, then allow. Global rules are checked before workspace rules within each tier.
- Three rule types -- Allow (auto-approve), Ask (prompt user), Veto (silently deny)
- Regex-based rules -- full JavaScript regex support, not limited to glob patterns
- Per-tool filtering -- scope rules to Bash, Read, Write, Edit, WebFetch, WebSearch, or any tool
- Global and workspace rules -- global rules apply everywhere, workspace rules are project-specific
- Live regex tester -- paste a sample command in the rule form to test before saving
- Activity Log -- see every intercepted request and whether it was auto-approved or passed through
- Persistent JSONL logs -- one log file per session, persisted to disk for auditing (click "Open Log Folder" in the Activity Log tab)
- Sound notification -- optional chime when a command needs manual approval (toggle in the panel header)
- Import/Export -- share rule sets as JSON
- Debug tab -- inspect raw extension activity and protocol messages
| Setting | Default | Description |
|---|---|---|
autoAuthorize.debug |
false |
Log all spawned processes and intercepted messages |
Auto-Authorize: Open Panel-- focus the sidebar panelAuto-Authorize: Enable-- enable auto-approvalAuto-Authorize: Disable-- disable auto-approval (all requests pass through)
THIS SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NONINFRINGEMENT.
This extension intercepts and modifies communication between VSCode and the Claude Code CLI by monkey-patching child_process.spawn. This is an unofficial tool, not affiliated with or endorsed by any company or individual. It may break at any time due to changes in Claude Code's internal protocol.
You use this extension entirely at your own risk. The author(s) accept no responsibility for any consequences resulting from its use, including but not limited to: unintended command execution, data loss, policy violations, or security issues. You are responsible for the regex rules you configure and the commands they approve.
BSD 3-Clause License. See LICENSE.txt for details.