The Agent Skills Directory

[COMMAND_EXECUTION]: Uses standard shell utilities including ls, grep, and git diff to analyze the local repository's structure and identify relevant documentation. These operations are restricted to read-only analysis as per the skill's invariants.
[EXTERNAL_DOWNLOADS]: Performs a version check using git ls-remote --tags <origin> to compare the local skill version with the upstream repository on GitHub. This is a benign network operation used for identifying available updates.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and summarize content from repository files, specifically searching for @ref annotations and system headers. While it processes untrusted data from the file system, the risk is mitigated by its strict read-only nature and focused purpose of citing existing documentation rather than executing logic based on the file content.

llp-orient