This release includes 1 breaking change for platform teams planning a safe upgrade.
✓ No known CVEs patched in this version
Topics
ReleasePort's take
Light signalThe release fixes a bug where PI chat responses repeated final assistant content when both streamed deltas and snapshots were present. It also introduces several new features such as AI PR cancellation, GPT‑5.6 preview handling, multi‑arch Docker images, and deprecates pinned markdown tables.
Why it matters: Fixes PI chat duplication (severity 40) affecting real‑time conversation flows; developers should verify toast handling after the refactor to avoid regressions.
Summary
AI summaryFixed PI chat responses repeating final content when both streamed deltas and snapshots were present.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Feature | Low |
Adds support for cancelling AI PR creation and passing backend choices into AI-assisted flows. Adds support for cancelling AI PR creation and passing backend choices into AI-assisted flows. Source: llm_adapter@2026-07-16 Confidence: high |
— |
| Feature | Low |
Adds Codex GPT-5.6 preview fast model handling. Adds Codex GPT-5.6 preview fast model handling. Source: llm_adapter@2026-07-16 Confidence: high |
— |
| Feature | Low |
Publishes multi‑arch server runtime Docker images with a slim runtime image. Publishes multi‑arch server runtime Docker images with a slim runtime image. Source: llm_adapter@2026-07-16 Confidence: high |
— |
| Deprecation | Low |
Removes pinned markdown table support, including the chat panel, persistence, and table pin actions. Removes pinned markdown table support, including the chat panel, persistence, and table pin actions. Source: llm_adapter@2026-07-16 Confidence: high |
— |
| Bugfix | Medium |
Fixes PI chat responses repeating final assistant content when streamed deltas and final snapshots were both present. Fixes PI chat responses repeating final assistant content when streamed deltas and final snapshots were both present. Source: llm_adapter@2026-07-16 Confidence: high |
— |
| Refactor | Low |
Simplifies chat PR toast handling and preserves PR resolver behavior after null checks. Simplifies chat PR toast handling and preserves PR resolver behavior after null checks. Source: llm_adapter@2026-07-16 Confidence: high |
— |
| Refactor | Low |
Keeps server Cargo package metadata in sync with desktop version bumps. Keeps server Cargo package metadata in sync with desktop version bumps. Source: llm_adapter@2026-07-16 Confidence: high |
— |
Full changelog
Features
- Added support for cancelling AI PR creation and passing backend choices into AI-assisted flows.
- Added Codex GPT-5.6 preview fast model handling.
- Published multi-arch server runtime Docker images with a slim runtime image.
Fixes
- Fixed PI chat responses repeating final assistant content when streamed deltas and final snapshots were both present.
Improvements
- Removed pinned markdown table support, including the chat panel, persistence, and table pin actions.
- Simplified chat PR toast handling and preserved PR resolver behavior after null checks.
- Kept server Cargo package metadata in sync with desktop version bumps.
Full Changelog: https://github.com/coollabsio/jean/compare/v0.1.63...v0.1.64
Breaking Changes
- Removed pinned markdown table support, including the chat panel, persistence, and table pin actions
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
Related context
Related tools
Beta — feedback welcome: [email protected]