The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to explain user-provided concepts and subsystems, which involves the agent ingesting and analyzing potentially untrusted source code from the user's project.
Ingestion points: The skill processes user-supplied concept names, subsystem descriptions, and project source code for the purpose of generating explanations.
Boundary markers: The instructions do not define specific delimiters for external content or include explicit warnings for the agent to ignore instructions embedded within the code being explained.
Capability inventory: The skill delegates investigation to other skills (how, why) and may use platform-specific image generation tools for creating diagrams. Its primary output is an explanation, and the instructions explicitly state that the skill should "change nothing" and has no side effects beyond read-only work.
Sanitization: No sanitization or filtering of the user-provided content or analyzed source code is specified in the instructions.
[EXTERNAL_DOWNLOADS]: The documentation and metadata reference external resources on GitHub, including the original pstack repository and other related forks. These links are provided for informational and attribution purposes and do not trigger automatic downloads or package installations during skill execution.

teach