Release 1.7.0
Release v1.7.0
This release improves the Claude PR review workflow with smarter, context-aware review depth, and corrects documentation around OAuth token authentication for the AI release workflow.
Features
Skill-First Review Depth in claude-pr-review
The PR review workflow now uses a skill-first approach to determine review depth:
- Skills are applied inline by default for standard diffs, keeping reviews fast and focused.
- Agents are automatically invoked for large or high-stakes diffs where deeper analysis is warranted.
This means you get lighter-weight reviews on routine changes and more thorough analysis where it matters, without any manual configuration changes required.
Bug Fixes
Corrected OAuth Token Documentation for ai-release
The documentation previously contained an incorrect claim about how OAuth tokens are passed. OAuth tokens are not accepted via x-api-key (requests will be rejected); they must be passed via the Bearer authorization scheme. If you were following the old docs and experiencing authentication failures, update your workflow configuration to use Bearer token auth.
Commits
- feat(claude-pr-review): skill-first review depth — skills inline by default, agents for large or high-stakes diffs (#43) (ecafe75)
- docs(ai-release): correct the OAuth-token claim — rejected via x-api-key, works via Bearer (#42) (529cb5d)
Full Changelog: v1.6.0...v1.7.0