The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's setup instructions involve cloning a repository from an untrusted source: https://github.com/0xBugatti/PentestOPS.git. This code is then executed in the user's environment via npm install and development server startups.
[INDIRECT_PROMPT_INJECTION]: The skill provides functionality to import external data from scanners and databases, which can be a vector for malicious content:
Ingestion points: Untrusted data enters the system through the Nmap XML import endpoint (/api/findings/import/nmap), CWE CSV imports (/api/cwes/import), and rich-text Editor.js content (/api/pages).
Boundary markers: There are no boundary markers or instructions to the agent to disregard embedded content within the imported XML or CSV files.
Capability inventory: The skill has the capability to write to the database (Finding.insertMany), perform file system operations (fs.promises.unlink), and generate reports (generateExecutiveSummary).
Sanitization: The provided code snippets do not show evidence of sanitization or validation for the content of the Nmap XML fields or CWE descriptions before they are processed and stored.
[COMMAND_EXECUTION]: The skill includes documentation for a maintenance script (backup.sh) that performs potentially destructive shell operations, including find -delete for file retention and docker exec for database operations.

pentestops-dashboard