v2026.916.0 v6.4.1
DavidHDev/react-bitsSep 16, 2026 → Sep 19, 20262 days between releases
What's New
Adds diagnostic skill, Native execution, and new coding harnesses like OpenCode 2.0
Breaking Changes
- `executing-plans` no longer stops every few tasks for review; it runs the whole plan then reviews once.
- Plans with the same basename no longer share a workspace.
New Features
- New `diagnosing-superpowers` skill figures out what went wrong in a session.
- Native execution mode for `executing-plans` offers a cheaper alternative to subagent-driven development.
- Support for new AI coding harnesses: OpenCode 2.0, Muse, and Qwen Code.
- Plans carry a Review Focus section to catch implied but untested failure modes.
Improvements
- Native execution mode is cheaper than subagent-driven development.
- Plans carry a Review Focus section to catch implied failure modes.
Upgrade Notes
Update `executing-plans` workflows to expect a single review at the end of the plan execution. Be aware that plans with identical basenames will now operate in separate workspaces to prevent conflicts.
The Two Releases
v2026.916.0
Sep 16, 2026· 1 month ago
A major release introducing AI credentials management via Connections, AgentMail for agents, and experimental chat connectors. The release also introduces GitHub App-backed identities and a complete Paperclip Runner execution engine.
Full release notesv6.4.1
Sep 19, 2026· 1 month ago
This release introduces a new diagnostic skill for debugging sessions, rebuilds plan execution to be cheaper via Native execution, and adds support for new AI coding harnesses like OpenCode 2.0, Muse, and Qwen Code.
Full release notes