## CHANGES
- Add SQLite datetime formats to version date parsing logic
- Loop through multiple date formats until one succeeds
- Include SQLite fractional seconds format support
- Prevent early return when version has PR numbers to output
- Simplify error handling for date parsing failures
- Remove duplicate database sync command from version workflow
- Simplify changelog generation to single process-prs step
- Keep database file staging after changelog generation
- Change last_pr_sync to use versionDate instead of time.Now()
- Ensure future runs fetch PRs merged after the version date
- Add clarifying comments explaining the sync timing logic
- Add cache metadata update step before staging release changes
- Set last_processed_tag to current version being processed
- Update last_pr_sync timestamp to current time
- Include warning messages for failed metadata updates
- Ensure metadata commits alongside other release changes
- Consolidate git worktree fixes into single PR #1917 entry
- Standardize commit message prefixes using conventional commits format
- Rewrite bullet points in imperative mood throughout changelog
- Condense verbose multi-line entries into concise single bullets
- Reorder PR entries chronologically within version sections
- Remove redundant Co-Authored-By attribution lines from entries
- Fix inconsistent date formats in version headers
- Simplify dependency update descriptions to essential information
- Update changelog database binary with new entry formatting
go-git has issues with worktrees where the object database isn't properly
shared, causing 'invalid object' errors when trying to commit. Switching
to native git CLI for add and commit operations resolves this.
This fixes generate_changelog failing in worktrees with errors like:
- 'cannot create empty commit: clean working tree'
- 'error: invalid object ... Error building trees'
Co-Authored-By: Warp <agent@warp.dev>
- Check filesystem existence of staged files to handle worktree scenarios
- Ignore files staged in main repo that don't exist in worktree
- Allow staged files that exist in worktree to be committed normally
Co-Authored-By: Warp <agent@warp.dev>
- Add worktree-specific check for actual working directory changes
- Filter out files that are only staged but not in worktree
- Check worktree status codes instead of using IsClean method
- Update GetStatusDetails to only include worktree-modified files
- Ignore unmodified and untracked files in clean check
- Extract message conversion into dedicated `toMessages` helper method
- Add proper role handling for system, user, and assistant messages
- Prepend system content to first user message per Anthropic format
- Enforce user/assistant message alternation with placeholder messages
- Skip empty messages during conversion processing
- Concatenate multiple text blocks in response output
- Add validation for empty message arrays before sending
- Handle edge case when only system content is provided
Add support for Google Cloud Vertex AI as a provider to access Claude models
using Application Default Credentials (ADC). This allows users to route their
Fabric requests through Google Cloud Platform instead of directly to Anthropic,
enabling billing through GCP.
Features:
- Support for Claude models (Sonnet 4.5, Opus 4.5, Haiku 4.5, etc.) via Vertex AI
- Uses Google ADC for authentication (no API keys required)
- Configurable project ID and region (defaults to 'global' for cost optimization)
- Full support for streaming and non-streaming requests
- Implements complete ai.Vendor interface
Configuration:
- VERTEXAI_PROJECT_ID: GCP project ID (required)
- VERTEXAI_REGION: Vertex AI region (optional, defaults to 'global')
Closes#1570
