The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes a vulnerability surface by directing the agent to ingest and follow penetration testing methodologies from an external repository.
Ingestion points: Numerous external Markdown files hosted on GitHub are listed in the 'Cheatsheet Library' section of SKILL.md.
Boundary markers: The skill lacks explicit delimiters or instructions to treat the external methodology text as data rather than instructions, potentially leading to the agent obeying commands embedded in the reference material.
Capability inventory: While SKILL.md does not define its own tools, the methodology describes a workflow utilizing reconnaissance and exploitation tools (e.g., nmap, amass, subfinder) in an environment expected to have the rea-universal-orchestrator active.
Sanitization: There are no mechanisms described for sanitizing or validating the content fetched from the external repository before it is processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill provides links to multiple external resources on GitHub (github.com/swisskyrepo/payloadsallthethings). While these are well-known documentation files for security research, they represent remote dependencies that the agent is encouraged to fetch during its workflow.

payloads-methodology