* feat: add Agent SDK support with USE_AGENT_SDK feature flag
Add a feature-flagged code path that uses the Agent SDK instead of
spawning the CLI as a subprocess. When USE_AGENT_SDK=true is set,
the new SDK path is used; otherwise, existing CLI behavior is unchanged.
Changes:
- Add parse-sdk-options.ts for parsing ClaudeOptions into SDK format
- Add run-claude-sdk.ts for SDK execution with query() function
- Update run-claude.ts with feature flag check at entry point
- Update update-comment-link.ts to handle both cost_usd and total_cost_usd
- Add @anthropic-ai/claude-agent-sdk dependency
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* refactor: simplify SDK types by using @anthropic-ai/claude-agent-sdk types directly
- Remove duplicate SdkRunOptions and McpStdioServerConfig types
- Use SDK's Options and McpStdioServerConfig types directly
- Return { sdkOptions, showFullOutput, hasJsonSchema } from parseSdkOptions
- Remove unnecessary convertMcpServers function
- Net reduction of ~70 lines
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* refactor: use extraArgs for claudeArgs pass-through to CLI
Simplify option parsing by converting claudeArgs to extraArgs record
and letting the SDK/CLI handle --mcp-config, --json-schema, etc.
- Remove extractJsonSchema and parseMcpConfigs functions
- Add parseClaudeArgsToExtraArgs for simple flag parsing
- CLI handles complex args like --mcp-config directly
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* ci
* refactor: remove hardcoded permission bypass flags
The SDK path should match CLI path behavior - permissions are handled
by the CLI itself, not hardcoded in the action.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* chore: add logging for SDK vs CLI path selection
---------
Co-authored-by: Claude <noreply@anthropic.com>
* refactor: update branch naming convention for Kubernetes compatibility
- Changed timestamp format in branch names to a shorter, Kubernetes-compatible style (lowercase, hyphens only).
- Updated related tests to reflect new branch name format.
- Ensured branch names are limited to a maximum of 50 characters to comply with Kubernetes naming requirements.
* refactor: clean up timestamp formatting in branch naming logic
- Removed unnecessary whitespace and standardized string formatting for the Kubernetes-compatible timestamp in branch names.
- Ensured consistency in the use of double quotes for string literals.
* feat: defer remote branch creation until first commit
- For commit signing: branches are created remotely by github-file-ops-server on first commit
- For non-signing: branches are created locally with 'git checkout -b' and pushed when needed
- Consolidated duplicate branch creation logic in github-file-ops-server into a shared helper function
- Claude is unaware of these implementation details and simply sees it's on the correct branch
- No branch links are shown in initial comments since branches don't exist remotely yet
* fix: prevent broken branch links in final comment update
- Check if branch exists remotely before adding branch link
- Only add branch links for branches that actually exist on GitHub
- Add test coverage for non-existent remote branches
- Fixes issue where users would see broken branch links for local-only branches
* fix: don't show branch name in comment header when branch doesn't exist remotely
- Only pass branchName to updateCommentBody when branchLink exists
- Prevents showing branch names for branches that only exist locally
- Add test to verify branch name is not shown when branch doesn't exist
* tmp
* feat: display detailed error messages when prepare step fails
- Capture prepare step errors in action.yml (up to 2000 chars)
- Add error details to comment update with collapsible section
- Handle both prepare and Claude execution failures separately
- Add test coverage for error detail display
This helps users debug issues like git errors, permission problems,
and branch creation failures more easily.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* refactor: simplify error capture to show clean error messages only
- Remove complex shell script that captured full output logs
- Use core.setOutput in prepare.ts to pass clean error message directly
- Avoid exposing potentially sensitive information from logs
- Show only the actual error message (e.g. 'Failed to fetch issue data')
This provides cleaner, more readable error messages without the risk
of exposing sensitive information from debug logs.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* refactor: simplify error display to show clean error messages only
- Remove collapsible <details> section for error messages
- Display errors in simple code blocks since messages are now clean and short
- Makes error messages more direct and readable
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
