Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
pyautogui
v1.2.0Cross-platform mouse/keyboard automation skill. Supports mouse control (move/click/drag/scroll), keyboard control (key press/hotkeys/type text), screen opera...
⭐ 0· 557·3 current·3 all-time
byIkaros@ikaros-521
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (mouse/keyboard automation, screenshots, image locating, cleanup) match the included scripts (keyboard_mouse.py, image_finder.py, draw_overlay.py, draw_on_image.py, cleanup.py). The packaged code implements the advertised features; no unrelated secrets or cloud credentials are requested.
Instruction Scope
SKILL.md gives concrete CLI commands that map to the included scripts. The instructions only reference installing Python packages and running the scripts; they do not instruct the agent to read unrelated system configuration or exfiltrate data. The cleanup script deletes files matching expected temporary-image patterns (subject to dry-run unless --execute is used).
Install Mechanism
There is no formal install spec in the registry (instruction-only), but the skill bundle contains multiple executable scripts. SKILL.md instructs installing Python packages via pip (pyautogui, Pillow, optionally opencv/rapidocr). This is consistent but worth noting: dependencies (especially RapidOCR/onnxruntime models) may download model data on first run and require additional disk/network usage; no arbitrary remote download URLs are embedded in the package itself.
Credentials
No environment variables, credentials, or config paths are requested. The skill's functionality (taking screenshots, controlling input, removing temporary image files) justifies the lack of declared secrets. There are no requests for unrelated tokens, keys, or system-wide config access.
Persistence & Privilege
always is false and the skill is user-invocable; it does not ask to run always or to modify other skills. The scripts perform local actions only (UI control, image processing, file deletion) and do not persistently modify agent/global settings.
Assessment
This package appears to be a straightforward PyAutoGUI-based automation toolkit and is internally consistent. Before installing/running it: 1) review and run the scripts in a safe environment (they can move the mouse, type, and delete files); use the cleanup script in preview mode first (--execute required to actually delete). 2) Be aware RapidOCR/onnxruntime may download or require large model files on first use and may use network/disk resources. 3) Granting accessibility/input permissions (macOS) or running with elevated rights (Windows) may be required for full functionality — avoid running with higher privileges than necessary. 4) If you plan to allow autonomous agent invocation, remember the skill can operate the UI—only enable autonomous use if you trust the skill and its source.Like a lobster shell, security has layers — review code before you run it.
latestvk972qvbdf9bwt08md1am815xxx82rk9a
License
MIT-0
Free to use, modify, and redistribute. No attribution required.