Skip to content

docs: changelog for v0.5.4 #365

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
dpup merged 1 commit into from
Jun 1, 2026
Merged

docs: changelog for v0.5.4 #365

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,10 +6,15 @@ Moat is pre-1.0. The CLI interface and `moat.yaml` schema may change between min

## Unreleased

## v0.5.4 — 2026-06-01

Patch release with three Claude Code fixes: an Ink-based TUI freeze on first paint, host/container session-directory divergence on workspaces with `.` or `_` in the path, and marketplace plugin hook scripts losing the executable bit.

### Fixed

- Fix Claude Code (and other Ink-based TUIs) freezing on the first paint inside the moat container — previously, when the child emitted any terminal-capability query that required a reply (CSI c Primary Device Attributes, CSI 6n cursor position, in-band resize, color queries), moat's VT compositor blocked forever on the emulator's internal reply pipe with no drain, holding `tui.Writer.mu` and starving the render goroutine. Input still reached the child, but no further output reached the host terminal, so the session appeared frozen and Ctrl+C wasn't visibly acknowledged even though it terminated the child. Surfaced 100% of the time with Claude Code 2.1.150+, which queries Device Attributes at startup. Moat now drains the emulator's reply pipe and routes replies back into the child's input stream via the existing injectable-reader chain. ([#362](https://github.com/majorcontext/moat/pull/362))
- Fix Claude Code sessions inside the container writing to a different `~/.claude/projects/` directory than host sessions — previously, moat slugified the workspace path by replacing only `/` with `-`, while Claude Code replaces every non-alphanumeric character. For any workspace path containing a `.`, `_`, or space (e.g. a macOS username like `user.name`), moat and the host CLI computed different project-directory names, so container and host sessions silently forked the project's session history and memory store. The slug now matches Claude Code's rule exactly (verified against the claude binary). Note: already-forked directories are not migrated automatically. ([#364](https://github.com/majorcontext/moat/pull/364))
- Fix marketplace plugin hook scripts failing with `Permission denied` inside the container — previously, `CollectMarketplaceTar` wrote every tar header with a hardcoded mode of `0644`, so executable scripts shipped by marketplace plugins (e.g. `bin/aw-hook`, `scripts/on-prompt-submit.sh`) extracted as non-executable and Claude Code's `UserPromptSubmit` hook failed at runtime. The tar header now derives `Mode` from the source file's permissions, preserving the executable bit. ([#363](https://github.com/majorcontext/moat/pull/363))

## v0.5.3 — 2026-05-25

Expand Down
Loading