The Agent Skills Directory

[PROMPT_INJECTION]: The skill demonstrates a surface for indirect prompt injection as it processes untrusted user input ('New Context') that is later merged into permanent knowledge assets. Malicious instructions embedded in this input could attempt to manipulate the agent's behavior during the diagnosis or execution phases.
[NO_CODE]: No executable scripts, binary files, or external software dependencies were found. The skill is entirely comprised of Markdown instructions and metadata, which minimizes the technical attack surface for command execution or remote code vulnerabilities.

asset-evolver