The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted external content from git diffs and specification files.\n\t- Ingestion points: The skill reads the output of git diff and git log commands, as well as the content of external specification files referenced in SKILL.md.\n\t- Boundary markers: The instructions do not define explicit delimiters or include warnings to ignore instructions embedded within the diff or spec files.\n\t- Capability inventory: The skill uses git commands (diff, log, rev-parse) and has the capability to write markdown reports to the filesystem.\n\t- Sanitization: No sanitization or validation of the ingested content is performed before processing.\n- [COMMAND_EXECUTION]: The skill executes shell-based git commands that incorporate user-supplied input for the git reference point.\n\t- Evidence: SKILL.md instructs the agent to run git diff <fixed-point>...HEAD and git log <fixed-point>..HEAD. While it performs a validation step with git rev-parse, the input flows directly into subsequent command executions.

unity-code-review