Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Psd Automator Screenshot
v1.0.1Use screenshot + natural language instruction to locate PSD text layers and dispatch automated edits with confidence gating. Requires psd-automator core.
⭐ 0· 237·1 current·1 all-time
by@dhrxy
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
The name/description (screenshot-driven PSD text edits) align with the SKILL.md usage examples. However the description says it "Requires psd-automator core" but the skill declares no dependency, no install, and no implementation files — it instead references built-in commands and a commandTool (psd_automator_screenshot) that are not provided or linked. That mismatch means the skill may silently fail or rely on other unlisted components.
Instruction Scope
The runtime instructions explicitly reference reading local files (e.g., ~/Desktop/修改3.png) and locating PSD files like banner.psd/banner.psb, which requires filesystem access to user files — expected for this purpose but not declared. The SKILL.md also states every invocation is recorded to skills-usage.json, which implies writing potentially sensitive usage data to disk. The doc’s "High confidence requests are auto-dispatched" behavior means the skill can perform edits without additional confirmations.
Install Mechanism
This is an instruction-only skill with no install spec and no code files, which minimizes direct install risk. There are no downloads or external install steps in the manifest.
Credentials
No environment variables, credentials, or config paths are requested (good). Still, the skill expects access to screenshots and PSD files on the user's filesystem and writes a skills-usage.json file — these I/O actions are reasonable for an editor skill but should be explicit to the user because they may include sensitive content.
Persistence & Privilege
always:false (no forced inclusion) which is appropriate. However autonomous dispatch of "high confidence" edits plus the skill's claim to record every invocation increases operational risk: combined with the agent's normal ability to invoke skills autonomously, this could result in edits and local logs without additional user confirmation. That is a behavioral risk, not a platform privilege mis-declaration.
What to consider before installing
What to consider before installing:
- Verify the dependency: ask where the "psd-automator core" lives and confirm it is installed and trusted; this skill references commands/tools that are not included.
- Confirm filesystem behavior: the skill expects you to pass screenshots and PSD filenames (e.g., ~/Desktop/…). Decide whether you are comfortable allowing the agent to read those files.
- Confirm logging: the skill records every invocation to skills-usage.json — ask where that file is stored and what data it contains (it may include instruction text or file paths).
- Auto-dispatch behavior: "High confidence requests are auto-dispatched" can cause edits without extra confirmation; if you want control, require that the agent ask for explicit confirmation before making changes.
- Source verification: there is no source or homepage; prefer skills with a repository or vendor you can review. If you proceed, test in a sandbox or with non-sensitive PSDs first and ask the publisher for implementation details and the storage location/format of skills-usage.json.Like a lobster shell, security has layers — review code before you run it.
latestvk97eb2502xtgcd7jycy5w56m3d83khpb
License
MIT-0
Free to use, modify, and redistribute. No attribution required.