v0.2.0 — Multi-provider + Streaming SSE

What's new in v0.2.0

App-side multi-provider support (Phase 3)

• New Translation Source picker: Direct API / Custom backend / 1st-party backend
• 7 direct providers — call APIs without a backend hop:
  • Gemini (Google AI Studio)
  • Ollama (local, privacy-first)
  • Google Translate Basic
  • OpenAI-compatible (LM Studio / OpenRouter / vLLM / OpenAI)
  • Gemini CLI (experimental)
  • Codex CLI (experimental)
  • Mock (echo, for smoke tests)
• Per-provider credential slots in macOS Keychain
• TranslationProviderFactory dispatches based on Settings; switch sources without re-entering keys

Streaming SSE (Phase 5)

• New /translate/stream endpoint emits SSE chunks
• Inbound flow shows progressive HUD as tokens arrive
• Mid-stream errors surface as RFC 7807 frames

v2 client adoption (Phase 2.5)

• Idempotency-Key on every request → server cache prevents double-paste
• RFC 7807 problem-details parsing → richer error messages
• 429 + Retry-After UX → "slow down, retry in 3s" instead of generic error

Production polish (Phase 7)

• X-Forwarded-For aware rate limiting (opt-in via env)
• DEPLOY guide with monitoring, troubleshooting, SSE nginx config

Tests

• 122 Swift / 29 suites + 68 pytest / 14 classes = 190 GREEN

Install (alpha)

1. Download `Contextual-Mac-Translator-v0.2.0-macos-arm64.dmg`
2. Open DMG → drag app to /Applications
3. Launch from /Applications
4. Grant Accessibility (System Settings → Privacy & Security)
5. Open Settings, pick a Source, paste credentials, test hotkey

Notes

• Apple Silicon only (arm64)
• Ad-hoc signed; first launch may need right-click → Open
• Bundle ID still placeholder (`app.lookerlab.translator`); Developer ID signing is on the M2 roadmap