- Add allowed_tools input to action.yml so workflows can specify allowed tools
- Pass allowed_tools to prepare step via environment variable
- Update agent mode to use allowed_tools when building claude_args
- Add GitHub file ops MCP server to agent mode when use_commit_signing is enabled
- This allows agent mode (used by auto-fix workflows) to write and commit files
- Add use_commit_signing: true to auto-fix workflows
- Update slash commands to use mcp__github_file_ops__commit_files
- This enables Claude to write and commit files through MCP
- Create new slash command specifically for when branch already exists
- Makes it crystal clear that Bash tool must be used for git commands
- Update workflow to use the new command
- Clarify that Claude must commit changes even when branch exists
- Update both slash command and inline workflow instructions
- Make git add and commit commands more explicit
- Add /fix-ci slash command for programmatic CI failure fixing
- Create auto-fix-ci.yml workflow using slash command approach
- Create auto-fix-ci-inline.yml workflow with full inline prompt
- Both workflows automatically analyze CI failures and create fix branches
The previous fix was looking in the wrong directory. Now uses CLAUDE_WORKING_DIR environment variable which points to the repository being processed, ensuring subagent files are found correctly.
- Reviews PR using embedded GitHub Actions variables
- Fetches PR details and diff using gh CLI
- Posts review as comment using gh pr comment
- Designed for use in automated workflows
- Multi-agent review process with parallel analysis
- Confidence scoring for issues (80+ threshold)
- Strict formatting for GitHub comments
- Uses gh CLI for all GitHub interactions
- Use exact same permissions as old workflow
- Use fetch-depth: 1 like old workflow
- Remove timeout and extra features
- Only trigger on opened PRs
- Use claude_args with --allowedTools format
- Add back review submission tools to allow Claude to comment
- Keep using v1-dev with the fixed agent mode
- This replaces the old claude-review.yml functionality
Pass MCP config and allowed tools through claude_args to ensure tools like
mcp__github_comment__update_claude_comment are properly available to Claude CLI.
Key changes:
- Tag mode outputs claude_args with MCP config (as JSON string) and allowed tools
- Fixed shell escaping vulnerability when JSON contains single quotes
- Agent mode passes through user-provided claude_args unchanged
- Re-added mcp_config input for users to provide custom MCP servers
- Cleaned up misleading comments and unused file operations
- Clarified test workflow is for fork testing
Security fix: Properly escape single quotes in MCP config JSON to prevent
shell injection vulnerabilities.
Co-Authored-By: Claude <noreply@anthropic.com>
- Add GitHub alert box highlighting the v1.0 roadmap
- Link to discussion #428 for community feedback
- Briefly summarize key changes (automatic mode selection, unified prompt interface)
- Position prominently at top of README for maximum visibility
The MCP servers (including github_comment server) were configured but not passed to Claude. This caused the "update_claude_comment" tool to be unavailable.
Changes:
- Write MCP config to a file at $RUNNER_TEMP/claude-mcp-config.json
- Add mcp_config_file output from prepare.ts
- Pass MCP config file via --mcp-config flag in claude_args
- Use fs/promises writeFile to match codebase conventions
- Agent mode now only triggers when explicit prompt is provided
- Removed automatic triggering for workflow_dispatch/schedule without prompt
- Re-added additional_permissions input for requesting GitHub permissions
- Fixed TypeScript types for mock context helpers to properly handle partial inputs
- Updated documentation to reflect simplified mode behavior
