The Agent Skills Directory

[SAFE]: The skill operates primarily as a retrieval-augmented guidance system for penetration testing. It contains comprehensive ground rules that prohibit unauthorized testing and mandate human verification at multiple stages of the workflow. All external links provided (pentesting.m14r41.in, checklist.m14r41.in) are legitimate vendor resources belonging to the author, and the skill maintains a professional, neutral tone throughout.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from external Markdown files to generate its output, which represents a potential attack surface. However, the risk is mitigated by explicit instructions to sanitize all outputs using placeholders (e.g., , ) and by requiring user confirmation before any testing steps. Mandatory Evidence Chain: 1. Ingestion points: Markdown files in the PentestingEverything repository; 2. Boundary markers: Explicit instructions to use placeholders and confirm targets/constraints; 3. Capability inventory: No direct tool access or network tools are requested in the YAML frontmatter; 4. Sanitization: Detailed instructions provided for replacing real secrets and targets with placeholders.
[COMMAND_EXECUTION]: The skill suggests security tool commands (e.g., curl, nmap, pdftotext) as part of its methodology. These are presented as educational templates with mandatory placeholders rather than commands for autonomous execution, ensuring the user remains in control of the assessment process.

pentesting-everything