The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local system commands including git diff, git status, and grep to identify and analyze code changes within the user's repository.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it ingests untrusted data from local files and git diff outputs into the agent's context without explicit sanitization.
Ingestion points: Local file content and git difference outputs are processed in SKILL.md (Stages 2 and 3).
Boundary markers: The workflow lacks specific delimiters or system instructions to ignore potential commands embedded within the code files being read.
Capability inventory: The skill employs git for repository metadata, read for file access, and grep for searching within files.
Sanitization: No evidence of content filtering or escaping of the ingested data was identified in the instructions.

yy-refresh