Features:
- Add Antigravity Provider for Google internal Gemini 3 Pro API
- Add dynamic route support (/:selector/v1/messages)
- Add protocol converter for Antigravity
- Add protocol selector for smart conversion path
Improvements:
- Add pre-commit hooks for local lint/format checks
- Simplify CI to only build check (lint moved to pre-commit)
- Update frontend with Antigravity support
Breaking Changes:
- Route parameter syntax changed from {param} to :param for Axum 0.7
- Fix health check request format with correct fields (chatTriggerType, modelId, origin)
- Load clientIdHash file correctly for IdC authentication
- Add success notifications for token refresh and health check
- Clean up debug println statements, use tracing instead
- Fix ESLint warnings
Version: 0.6.1
- Auto start server on app launch
- Fix token usage statistics (estimate from content length)
- Add token expiry check before requests (10 min threshold)
- Change default port to 8999, api_key to proxy_cast
- Fix VecDeque usage in logger
- Fix save_credentials await in providers
- Fix tool_use input parsing: use HashMap to track concurrent tool calls
- Correctly accumulate input fragments for each tool_use event
- Fix tool_result conversion in openai_to_cw.rs
- Add streaming response text block for Claude Code compatibility
- Add model mappings for claude-opus-4-5-20251101 and claude-haiku-4-5-20251001
- Add debug logging for CodeWhisperer requests
- Add unified version management script (scripts/update-version.sh)
- Apply cargo fmt and prettier formatting
- Add SSE streaming response for Anthropic /v1/messages endpoint
- Add tool_calls support in response parsing
- Add detailed request/response logging for debugging
- Enhance API compatibility check with tool_call test
- Add Claude Code compatibility check UI in Settings
- Add check_api_compatibility Tauri command
- Add API compatibility check UI in Settings page
- Test all configured models with health check
- Show last successful check time
- Display warnings for 401/403/400 errors
- Fix clippy format string warnings
- Add Default trait implementations for all providers
- Use inline format string variables (clippy::uninlined_format_args)
- Use Range::contains for status code checks
- Fix needless_range_loop in openai_to_cw converter
- Run cargo fmt to fix formatting
- Add GitHub Actions CI workflow with:
- Rust clippy lint check
- Rust format check
- Frontend TypeScript check
- Build verification
- Add OpenAICustomProvider for custom OpenAI-compatible APIs
- Add ClaudeCustomProvider for custom Claude APIs
- Add configuration UI with API key and base URL inputs
- Update Providers page with all 5 provider tabs
- All providers now manageable from single page
- Add GeminiProvider with OAuth token refresh support
- Add Gemini credential management commands
- Update Providers page with tabs for Kiro/Gemini
- Auto-detect credential file changes for both providers
- Add Gemini models to /v1/models endpoint
- OAuth credentials must be set via GEMINI_OAUTH_CLIENT_ID/SECRET env vars
coso