The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard Unix utilities including find, sed, grep, and comm to compare the local filesystem against the documentation index. It also includes a local verification step using pnpm test within the project's own CLI package. These operations are legitimate, limited to the local environment, and align with the skill's purpose of documentation auditing.
[PROMPT_INJECTION]: The skill processes markdown (.mdx) files from the local repository during the auditing phase. While reading untrusted documentation content can theoretically serve as a vector for indirect prompt injection, the skill does not perform high-risk operations or grant elevated privileges that would make this surface exploitable in a malicious way. The behavior is consistent with standard documentation management tasks.

phoenix-llms-txt