Skip to content

ChaosEngineAI v0.7.6

Choose a tag to compare

View all tags
@github-actions github-actions released this 09 May 09:50
· 289 commits to staging since this release
v0.7.6
f86ac4e
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

v0.7.6 — HTML Challenge + UI scale + retry

HTML Challenge (the headline)

New chat sub-tab dedicated to side-by-side HTML generation. Pick 2 → 4 model slots, each with its own:

  • Model + full launch settings (engine, cache strategy, KV bits, context length)
  • Thinking mode (off / auto)
  • Reasoning effort (low / medium / high)
  • Seed

Issue one prompt; every slot streams its HTML response in parallel with a sandboxed live preview under the raw text — so you see what each model actually rendered, not just what it wrote.

  • Automated HTML validation per slot — valid / partial / script-error / blank-render / no-html. Fed back from the iframe sandbox so script crashes and blank renders surface immediately.
  • Persistent history — every challenge run is saved with title + prompt + per-slot manifests (model, settings, thinking mode, reasoning effort, seed, validation status, output path). Re-open from the history view, delete, or open / reveal individual slot HTML files on disk.
  • Retry + repair flows — re-run a single failed slot, ask the model to continue from the partial response, or kick off a repair pass that asks the same model to fix its own broken HTML.
  • New routes under /api/chat/html-challenges (SSE-streamed run, history list, fetch one, delete one, fetch slot file, retry slot, repair slot, patch validation, open file).

UI scale

Segmented UI scale control in Settings → Display: 75% / 100% / 125% / 150%. Applies app-wide via a useUiScale hook so the entire workspace re-flows around the chosen density. Persisted across launches.

Chat polish

  • Retry failed generation — failed assistant messages now expose a Retry button; one click re-issues the previous turn against the active runtime profile without re-typing.
  • Chat draft state hardened — drafts persist per thread across tab switches.
  • Reasoning panel: streaming preview height + collapse animation fixes.
  • Runtime controls: tighter spacing + clearer cache strategy / DFlash gating.

Compare view refactor

CompareView modularized so the two-up Compare mode and the new HTML Challenge multi-slot view share buildComparePayload / cloneLaunchSettings / compareTargetLabels / useLaunchPreview instead of duplicating slot configuration logic.

Gallery flows

Image + Video Gallery: filter / sort / re-run flow tightened. Video gallery now mirrors image-gallery's metadata layout + reveal-on-disk + clone-settings actions.

Packaging

  • Application version bumped to 0.7.6 across the npm, Python, and Tauri package metadata.
  • v0.7.5 was used internally during HTML Challenge development; superseded directly by 0.7.6.

Bundles below: macOS aarch64 dmg + Linux x64 AppImage / deb + Windows x64 setup.exe. latest.json for Tauri auto-updater.

Assets 13
Loading