Smart Skill Finder
Analysis
The skill’s purpose is coherent, but it needs review because its code passes user search text into a shell command and may overstate security verification of recommended skills.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.
search_terms = " ".join(query.get('keywords', [])); cmd = f'npx skills find "{search_terms}" --json'; subprocess.run(cmd, shell=True, capture_output=True, text=True, timeout=10)Search terms derived from the user's natural-language query are interpolated into a shell command and run with shell=True. Quoting alone does not prevent a crafted quote or shell metacharacter from changing the command.
install_command=f"npx skills add {skill_data.get('package', '')}", ... security_status='clean'The Skills CLI parser assigns a clean security status while building an install command, without showing scanner evidence in that branch. The presentation code later maps clean status to a security-verified message.
"default_install_commands": { "skills_cli": "npx skills add {package}", "clawhub": "clawhub install {author}/{skill}", "github": "git clone {repo_url} ~/.openclaw/skills/{skill_name}" }The skill’s core function is to recommend installing third-party skills from multiple ecosystems. This is disclosed and user-directed, but it is inherently supply-chain-sensitive.
Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.
- Reads your existing skills to avoid duplicates - Considers your project context and domain - Can reference past skill discovery conversations - Builds knowledge of your preferred ecosystems
The README describes using workspace context and past conversations/preferences. That fits personalized recommendations, but the artifacts do not define retention, scope, or opt-out behavior.