Image2Tools v0.2.1 (unsigned preview)

Image2Tools v0.2.1 (unsigned preview)

Bugfix and UX release based on post-0.2.0 user feedback.

Changes

• Reference-image editing and inpaint requests now run without SSE streaming, fixing aggregator/proxy failures such as provider returned non-SSE response for streaming request on /images/edits.
• The provider dropdown has been removed. The app preserves saved configuration and relies on model discovery to infer provider/model availability.
• The workspace mode tabs are simplified to Text to image and Image to image. Masks are now optional under Image to image while internal generate/edit/inpaint routing remains intact.
• Draft restore, history reuse, Gemini guided-region wording, and General fallback behavior remain compatible with the v0.2.0 multi-model workflow.

Validation

• PR #113 CI passed: build/mock verifier, macOS package, Windows package, Linux package.
• Local release validation passed on macOS arm64:
  • pnpm install --frozen-lockfile
  • pnpm build
  • pnpm verify:mock-api
  • pnpm verify:mock-gemini-api
  • pnpm verify:mock-model-discovery
  • pnpm verify:release-evidence
  • pnpm package:mac
  • pnpm verify:release:mac

Release Status

This is an unsigned/ad-hoc signed preview prerelease. macOS artifacts are not Apple-notarized, so Gatekeeper may warn on first launch. Real OpenAI/Gemini API acceptance and Developer ID signing/notarization remain pending in docs/release/evidence.json.

What's Changed

• v0.2.1: streaming-edit fix, provider auto-detect, two-tab workspace by @Bliveren in #113

Full Changelog: v0.2.0...v0.2.1

What's Changed

• v0.2.1: streaming-edit fix, provider auto-detect, two-tab workspace by @Bliveren in #113

Full Changelog: v0.2.0...v0.2.1