Skip to main content

skills-audit

Conducts comprehensive security audits of AI Agent skills to identify vulnerabilities and ensure safe deployment.

Install this skill

or
0/100

Security score

The skills-audit skill was audited on May 12, 2026 and we found 24 security issues across 5 threat categories, including 8 critical. Review the findings below before installing.

Categories Tested

Security Issues

critical line 56

Direct command execution function call

SourceSKILL.md
56- **Remote Code Execution**: `eval()`, `exec()`, `subprocess`, `curl | bash`
critical line 202

Direct command execution function call

SourceSKILL.md
202- **Remote Code Execution**: `curl | bash`, `eval()`, `exec()`
critical line 56

Eval function call - arbitrary code execution

SourceSKILL.md
56- **Remote Code Execution**: `eval()`, `exec()`, `subprocess`, `curl | bash`
critical line 202

Eval function call - arbitrary code execution

SourceSKILL.md
202- **Remote Code Execution**: `curl | bash`, `eval()`, `exec()`
medium line 21

Template literal with variable interpolation in command context

SourceSKILL.md
21```bash
critical line 56

Piping content to bash shell

SourceSKILL.md
56- **Remote Code Execution**: `eval()`, `exec()`, `subprocess`, `curl | bash`
high line 117

Piping content to bash shell

SourceSKILL.md
117"code_snippet": "echo 'L2Jpbi9iYXNoIC1jIC...' | base64 -D | bash",
critical line 202

Piping content to bash shell

SourceSKILL.md
202- **Remote Code Execution**: `curl | bash`, `eval()`, `exec()`
medium line 136

Webhook reference - potential data exfiltration

SourceSKILL.md
136- **Webhook is NOT sent during this step** -- it will be sent after your comprehensive analysis
medium line 138

Webhook reference - potential data exfiltration

SourceSKILL.md
1383. **Send final webhook notification** (optional, if notifications are configured)
medium line 139

Webhook reference - potential data exfiltration

SourceSKILL.md
139After completing comprehensive analysis (including false positive filtering), send the webhook:
low line 145

Webhook reference - potential data exfiltration

SourceSKILL.md
145from skill_audit.integrations import send_final_webhook
low line 146

Webhook reference - potential data exfiltration

SourceSKILL.md
146send_final_webhook(report_path='<report_path>')
medium line 149

Webhook reference - potential data exfiltration

SourceSKILL.md
149This ensures the webhook contains the final, accurate results after your analysis.
medium line 256

Webhook reference - potential data exfiltration

SourceSKILL.md
256- **Webhook is deferred** until after Claude's comprehensive analysis (false positive filtering)
low line 25

Access to hidden dotfiles in home directory

SourceSKILL.md
25~/.claude/skills/skills-audit/skill_audit/cli_wrapper.py \
low line 26

Access to hidden dotfiles in home directory

SourceSKILL.md
26~/.claude/skills/skill-audit/skill_audit/cli_wrapper.py \
low line 220

Access to hidden dotfiles in home directory

SourceSKILL.md
220custom_report_dir: ~/.claude/audit-reports
medium line 253

Access to hidden dotfiles in home directory

SourceSKILL.md
253- **Reports saved to ~/.claude/audit-reports/** by default (configurable)
medium line 57

Access to .env file

SourceSKILL.md
57- **Credential Leaks**: Hardcoded API keys, passwords, tokens, .env files
medium line 203

Access to .env file

SourceSKILL.md
203- **Credential Leaks**: Hardcoded API keys, passwords, .env files
low line 119

Base64 decode operation

SourceSKILL.md
119"scenario": "User follows installation instructions, base64 decodes to malicious payload, executes with shell privileges",
critical line 59

Prompt injection: ignore instructions

SourceSKILL.md
59- **Prompt Injection**: "Ignore previous instructions", role manipulation
critical line 206

Prompt injection: ignore instructions

SourceSKILL.md
206- **Prompt Injection**: "Ignore previous instructions"
Scanned on May 12, 2026
View Security Dashboard
Installation guide →
GitHub Stars 325
Rate this skill
Categorydevelopment
UpdatedMay 13, 2026
xwtro0tk1t-cloud/harness