pentest-lyan-web-security-testing
Warn
Audited by Gen Agent Trust Hub on Jun 27, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to clone code from a repository at 'https://github.com/HeaSec/Pentest-Lyan.git', which is not a verified or trusted source.
- [CREDENTIALS_UNSAFE]: The skill explicitly documents that it stores project credentials (including usernames and passwords) in plaintext within the 'pentest-data/' directory. While it advises adding this directory to '.gitignore', the storage of unencrypted secrets on the local filesystem is a significant security risk.
- [PROMPT_INJECTION]: The skill is highly susceptible to indirect prompt injection. It autonomously crawls, downloads, and processes untrusted external content (such as JavaScript bundles and API responses) from target websites to perform discovery and attack phases.
- Ingestion points: Processes external JavaScript via 'extract_signing_function' and evaluates API responses during 'build_permission_matrix' generation.
- Boundary markers: No delimiters or 'ignore embedded instructions' warnings are present to protect the agent from malicious instructions hidden in target website data.
- Capability inventory: The skill possesses broad capabilities including file system writes ('pentest-data/'), network requests ('requests' library), and execution of utility scripts ('render_docx.py').
- Sanitization: There is no evidence of sanitization or safety filtering for the data ingested from target web applications.
- [COMMAND_EXECUTION]: The installation process involves global package installations ('npm install -g @playwright/mcp-server') and the execution of Python scripts ('scripts/render_docx.py') which could facilitate privilege escalation or arbitrary code execution if the downloaded repository is compromised.
Audit Metadata