cua-driver

Drive a native GUI app (macOS, Windows, Linux) via the Qwen Cua Driver CLI (default) or MCP server; snapshot its accessibility tree, act through snapshot-bound element tokens, native menu paths, exact window geometry, or pixel coordinates, and verify from fresh state. Use when the user asks you to operate, drive, automate, or perform a GUI task in a real application on the host.

Quality

90%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

Quality

Content

81%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A dense, highly actionable operational contract with an exemplary verify-then-escalate workflow and concrete code throughout. Its main weakness is conciseness: the invariant is restated several times and some sections (cursor overlay cosmetics, historical reproducers) pad an already long file.

Suggestions

State the snapshot-before-action / verify-after invariant once in 'The core invariant' section and reference it elsewhere instead of re-deriving it in the intro, the canonical loop, and 'Verify after every action'.

Move the agent-cursor-overlay theme/animation detail (glow, action marks, session chips) into a dedicated companion file or trim it to the one operational caveat (AX-run seeding) that actually affects driving.

Ship the referenced companion files (MACOS.md, WINDOWS.md, LINUX.md, BROWSER.md, RECORDING.md) in the bundle so the one-level-deep navigation the body promises actually resolves.

Dimension	Reasoning	Score
Conciseness	The body is mostly operational and novel (driver-specific contract Claude does not already know), but at ~1050 lines it repeats the snapshot-before-action invariant in four places and includes a long cursor-overlay cosmetics section and a historical IINA reproducer that could be trimmed.	3 / 5
Actionability	Provides copy-paste-ready bash CLI invocations with exact JSON args, a Python crosshair snippet, and a tool dispatch table with precise signatures covering the common cases — fully executable guidance.	5 / 5
Workflow Clarity	The verify-then-escalate ladder is an explicit numbered algorithm with a validation checkpoint after every action ('verify after every action — mandatory'), feedback loops (verify → fix → re-snapshot → retry), and clear escalation rungs.	5 / 5
Progressive Disclosure	Section headers are clear and companion files (MACOS/WINDOWS/LINUX/BROWSER/RECORDING.md) are signaled one level deep, but no bundle files are actually present in the directory and the core SKILL.md is a long monolith inlining detail that the structure implies lives elsewhere.	4 / 5
	Total	17 / 20 Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A tightly written third-person description that names the domain, lists several concrete capabilities, and gives an explicit 'Use when' trigger with natural synonym coverage. It does not pad or over-claim.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'snapshot its accessibility tree', 'act through snapshot-bound element tokens, native menu paths, exact window geometry, or pixel coordinates', 'verify from fresh state' — giving comprehensive coverage of the skill's capabilities.	5 / 5
Completeness	Explicitly answers both what ('Drive a native GUI app ... snapshot ... act ... verify from fresh state') and when ('Use when the user asks you to operate, drive, automate, or perform a GUI task in a real application on the host') with concrete trigger phrases.	5 / 5
Trigger Term Quality	The 'Use when' clause supplies natural synonyms users would actually say — 'operate, drive, automate, or perform a GUI task' — alongside 'native GUI app' and the three OS names; file extensions are not applicable to this skill.	5 / 5
Distinctiveness Conflict Risk	Occupies a clear niche — native GUI automation via the Qwen Cua Driver — with distinct triggers (operate/drive/automate a real host application) and minimal overlap risk with other skills.	5 / 5
	Total	20 / 20 Passed

Validation

75%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 12 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (1082 lines); consider splitting into references/ and linking	Warning
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	12 / 16 Passed

Repository: QwenLM/qwen-code
Path: packages/cua-driver/rust/Skills/cua-driver/SKILL.md
Commit: 054eabb

Reviewed: 1 day ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.