The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill utilizes dynamic context injection (!) to execute git ls-files and eza . --tree at load time. These commands are standard directory and file listing tools used to provide the agent with codebase structure context, which is consistent with the skill's primary purpose. No sensitive data access or network exfiltration is performed via these injections.
[INDIRECT_PROMPT_INJECTION]: The skill ingests content from local files such as README.md and documentation within the ai_docs/ directory. This creates an ingestion surface for potentially untrusted data from the codebase, but there are no high-privilege capabilities exposed that would allow for exploitation via this surface.

prime