chore(cua-driver-rs): drop -rs from MCP identity + skill pack name - #1677
Merged
Conversation
Three user-visible places where the binary still advertised "-rs" after the libs/cua-driver-rs/ folder rename. Renamed all three and added one-shot legacy cleanup so existing installs migrate silently. 1. MCP `serverInfo.name`: "cua-driver-rs" → "cua-driver" (mcp-server/src/protocol.rs) 2. MCP `instructions` text: "cua-driver-rs: cross-platform…" → "cua-driver: …" and the in-string skill-name reference too. This is the text Claude / Codex / etc. see in their system prompt. 3. SKILL_PACK_NAME: "cua-driver-rs" → "cua-driver" + repo dir git-mv libs/cua-driver/rust/Skills/cua-driver-rs/ → cua-driver/. Install dir on disk changes from <HomeDir>/skills/cua-driver-rs/ to <HomeDir>/skills/cua-driver/, and same for agent links under ~/.claude/skills/, ~/.agents/skills/, ~/.openclaw/skills/, and the OpenCode dir. Migration (zero-prompt, runs on next `cua-driver skills install`): - new `sweep_legacy_skill_pack()` removes <HomeDir>/skills/cua-driver-rs/ and any `<agent_skills>/cua-driver-rs` symlink/junction (only when it's a managed link — never clobbers a user-managed real dir). - `skills uninstall` now sweeps BOTH the current name and the legacy one so a partial uninstall still finishes. - uninstall.sh + uninstall.ps1 expanded their junction lists to include both names. Asset name `cua-driver-rs-v<v>-skills.tar.gz` intentionally unchanged per the broader "tag prefix + asset names stay" decision. The CD workflow's Skills staging now packages a `cua-driver/` top-level dir inside the (legacy-named) tarball; the on-binary tarball extractor strips whatever top-level dir it finds, so both old and new tarballs extract correctly. Smoke-tested on Windows: `cua-driver mcp` initialize returns `serverInfo.name = "cua-driver"` and instructions text starts with `"cua-driver: cross-platform…"`. cargo build/test all green.
Contributor
|
The latest updates on your projects. Learn more about Vercel for GitHub. |
Contributor
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (14)
📝 WalkthroughWalkthroughThis PR renames the cua-driver skill-pack from ChangesSkill-pack rename and backwards-compatible migration
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
4 tasks
f-trycua
added a commit
that referenced
this pull request
May 24, 2026
… .cua-driver-rs/ + repairs dangling agent links (#1683) User reported `cua-driver skills install` placing the skill pack at `~/.cua-driver-rs/skills/cua-driver/` even after PR #1677 renamed the SKILL_PACK_NAME — because skills.rs's home_dir() still hardcoded `.cua-driver-rs/` as the package-home subdir, inconsistent with telemetry.rs which migrated to `.cua-driver/` in v0.2.16. Three changes: 1. **home_dir() now defaults to `~/.cua-driver/`** (lifted the same HOME_SUBDIRECTORY / LEGACY_HOME_SUBDIRECTORY constants telemetry.rs already uses). `CUA_DRIVER_RS_HOME` env override still wins. 2. **sweep_legacy_skill_pack() expanded** to clean any of three legacy shapes: - <NewHome>/skills/cua-driver-rs/ (old pack NAME under new home) - <LegacyHome>/skills/cua-driver/ (new pack NAME under old home — the user's case) - <LegacyHome>/skills/cua-driver-rs/ (old NAME under old home) Then attempts to remove the empty `<LegacyHome>/skills/` and `<LegacyHome>/` themselves — but only if they're actually empty (`fs::remove_dir`, never `remove_dir_all`), so a legacy Unix install that still has a packages/ tree alongside keeps its dot-folder. 3. **link_agent() repairs dangling links**, not just creates missing ones. After sweep removes the old target, the agent dir (`~/.claude/skills/cua-driver` etc.) holds a junction pointing at nothing. The old guard `if link.exists() || symlink_metadata.is_ok` misread that as "already linked, skip" and left it broken. Now: - link.exists() = true → real / valid link (skip) - has metadata + !exists + is_symlink/junction → dangling, remove + recreate - real directory → user-managed (skip) `is_symlink_or_junction` already gated dangerous removals — never touches a user-managed real dir. `uninstall --all` similarly extended to sweep the legacy home dir. End-to-end verified on Windows: - legacy pack at ~/.cua-driver-rs/skills/cua-driver ← gone after install - stale junction at ~/.claude/skills/cua-driver ← repaired (now points at ~/.cua-driver/skills/cua-driver) - new home ~/.cua-driver-rs/ ← removed (was empty after sweep)
Merged
5 tasks
f-trycua
added a commit
that referenced
this pull request
May 24, 2026
…apper from skill tarball (#1684) User reported `~/.claude/skills/cua-driver/cua-driver-rs/SKILL.md` after `skills install` on Windows — a nested `cua-driver-rs/` dir inside the pack root. Root cause: tarball-shape mismatch with the extractor. The CD workflow historically staged files at `<outer>/cua-driver-rs/` (and after #1677, `<outer>/cua-driver/`) — TWO wrapping dirs. The extractor only stripped ONE. So files landed at `dest/cua-driver{-rs}/<file>` instead of `dest/<file>`. Two pieces: 1. **`extract_tar_gz` now strips a second wrapper IF named `cua-driver` or `cua-driver-rs`**. This covers three historical tarball shapes: - v0.2.18 and earlier: `…-skills/cua-driver-rs/<file>` - v0.2.19 (briefly): `…-skills/cua-driver/<file>` - v0.2.20+ (post-fix): `…-skills/<file>` (flat — CD workflow now) Stripping is name-gated so a future skill pack with a real subdir (e.g. `examples/`) doesn't get accidentally flattened. 2. **CD workflow flattens the staging dir** — no more redundant inner `cua-driver/` between `<outer>` and the .md files. New tarballs are single-wrap. Four unit tests cover the three legacy shapes + a "preserve real subdir" guard. End-to-end fix: on the user's machine, the live v0.2.18 release tarball still has the legacy double-wrap shape, so the binary needs to keep extracting it correctly via the smart extractor — which it now does. `skills update` will sweep the nested `cua-driver-rs/` dir out (`fetch_into` wipes dest first) and re-extract flat.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Three user-visible places where the binary still advertised `-rs` after the `libs/cua-driver-rs/` folder rename. Renamed all three and added one-shot legacy cleanup so existing installs migrate silently.
What changed
The `serverInfo.name` + `instructions` change is what fixes the other-Claude-session complaint that prompted this PR ("the cua-driver-rs skill isn't loaded").
Migration (zero-prompt, runs on next `cua-driver skills install`)
What's intentionally NOT changing
Per the user-confirmed scoping from PR #1674:
Test plan
🤖 Generated with Claude Code
Summary by CodeRabbit
Bug Fixes
Documentation