v0.5.0: Phase 4 Parallel Suggestion Generation

Phase 4: Parallel Suggestion Generation

Key Features

• Parallel Streaming: Implement concurrent suggestion generation with concurrency limit of 10, achieving ~10x speedup (from ~100s to ~10s for 20 terms)
• Progress Tracking: Real-time progress indicator showing "Generating X/Y..." during generation
• Local Development Support: Fixed local development to work without Secrets Store access

Implementation Details

Parallel Streaming Utility

• New parallelStream() async generator in packages/api/src/platform/parallel.ts
• Respects concurrency limits while yielding results immediately as tasks complete
• Supports error isolation using PromiseSettledResult

Suggestion Generation Updates

• Updated generateSuggestions() to use parallel streaming instead of sequential processing
• Results stream to client as each LLM call completes for responsive UI

Progress Tracking

• Enhanced BatchDetailPage and BatchHeader components
• Tracks completion progress with ref-based guard to prevent duplicate API calls
• Displays precise progress feedback to users

Local Development Improvements

• Added getSecretValue() helper supporting dual-mode secrets
• Cloudflare Secrets Store bindings now production-only
• Local .dev.vars seamlessly supports development workflows

What's Fixed

• Local development environment now works independently of remote Secrets Store access
• Removed unused AI binding that caused connection issues

Commits Included

• 4141809: feat(api,web): implement parallel suggestion generation (Phase 4)
• c2a6db5: fix(api): support local dev without Secrets Store access

Related Issues

Closes implementation of Phase 4 from project roadmap