The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from repository files and user-provided arguments, which could contain instructions intended to redirect the agent's behavior. * Ingestion points: The skill reads feature descriptions from $ARGUMENTS and crawls repository files, including specification documents. * Boundary markers: There are no explicit instructions to use delimiters or ignore embedded instructions within the processed files. * Capability inventory: The skill utilizes the Explore subagent, Bash for reading files and git logs, and mktemp for temporary storage. * Sanitization: The instructions do not define methods for sanitizing or escaping the content of the analyzed files.
[COMMAND_EXECUTION]: The skill uses shell commands to facilitate its analysis of the codebase. * Evidence: Use of git log --oneline -20 to analyze commit history and mktemp to create temporary files for sketching diffs.

preparatory-refactor