perf-auditor
Verdict safe, but 1 finding (1 medium), mainly social engineering. Most recent scan 2026-02-06.
This skill package has MEDIUM security concerns primarily due to deceptive presentation. The skill masquerades as an agent skill but actually provides documentation for an external npm package (ai-lighthouse) without including any executable scripts. This creates a significant mismatch between user expectations and actual functionality. The skill is not inherently malicious - it does not attempt data exfiltration, prompt injection, or credential theft. However, the lack of actual implementation code combined with instructions to run external tools (npx) creates ambiguity about what the agent would actually execute. The missing manifest metadata (license, compatibility, allowed-tools) and embedded promotional content are minor issues. If this skill were to include actual scripts that execute 'npx ai-lighthouse', additional security analysis would be needed to assess the risks of running external npm packages.
Scanner's overall assessment (LLM analyzer)
Second scanner: NVIDIA SkillSpector
Independent static run, no language model, over the same files as the Cisco scan above. How the two scanners compare across the whole corpus: Two scanners, one corpus.
What this skill does
Run a Lighthouse performance audit with AI fix suggestions. Use when your site is slow and you need actionable fixes.
- Package
- 2 files, 3.0 KB
- SKILL.md
- 391 words, 12+ sections
- Code
- None. Instructions only.
SKILL.md covers: Performance Auditor · Quick Start · What It Does · Usage Examples · Audit a production URL · Audit a local dev server · Audit a specific page · Best Practices · When to Use This · Part of the LXGIC Dev Toolkit · Requirements · How It Works
Findings
MEDIUM Deceptive Skill Description - External Tool Masquerading as Agent Skill Social Engineeringllm
The skill description and instructions describe an external npm package (npx ai-lighthouse) rather than an actual agent skill with executable code. The SKILL.md contains only documentation for running an external tool via 'npx ai-lighthouse', with no Python/Bash scripts provided. This creates a mismatch between what users expect (an agent skill that performs audits) and what is delivered (instructions to run an external npm package). Users may be misled into thinking this is a self-contained skill when it actually requires external package execution.
SKILL.md
Quick Start section shows: 'npx ai-lighthouse https://mysite.com' - This is an external npm package call, not agent skill functionality. No Python/Bash scripts are included to implement the described functionality.What is in the package
| File | Kind | Size |
|---|---|---|
README.md | doc | 453 B |
SKILL.md | doc | 2.6 KB |
External hosts referenced
mysite.com github.com lxgic.dev x.com lxgicstudios.substack.com
Hostnames that appear in the skill's files. Referencing a host is not a finding; compare against the findings above and the skill's stated purpose.
Package contents recorded 2026-10-22. The scan above ran against the version current at scan time; the two can differ if the skill has since been updated.
What we checked
- Static analyzer. Pattern rules over SKILL.md and bundled files: known-bad commands, exfiltration URLs, secrets, obfuscation.
- Behavioral analyzer. Checks what the skill asks the agent to do against what its description claims it does.
- Llm analyzer. A language model reads the skill and reports intent-level issues the rules cannot express.
- Trigger analyzer. Looks for instructions that fire on conditions the user did not ask for.
- Meta analyzer. A second model pass that validates, deduplicates and re-scores the other analyzers' findings.
What we did not check
- Runtime behaviour. The skill was not executed; findings come from reading its files.
- Anything fetched at install or run time, including remote scripts, package dependencies, and URLs the skill tells the agent to open.
- Content the analyzers cannot read: images, binaries, and some non-English text.
- Known scanner bypass classes. See the methodology and limitations page.
Sources
- Repository: https://clawhub.ai/skills/perf-auditor
- Raw scan result: performance-auditor-scan.json
- Badge for your README:
- Request a re-scan