The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes arbitrary code changes from the local environment to generate its output.
Ingestion points: The skill uses git diff --staged to ingest code changes into the agent's context for analysis (SKILL.md).
Boundary markers: There are no defined delimiters or instructions provided to the model to ignore potential natural language instructions embedded within the code diffs (e.g., in comments).
Capability inventory: The skill has the capability to analyze files and presumably execute git commands via referenced helper scripts (SKILL.md).
Sanitization: No sanitization or filtering of the ingested diff data is mentioned or performed within the instruction file.
[COMMAND_EXECUTION]: The skill requires the execution of shell commands to interact with the Git version control system.
Evidence: The "Analysis Flow" section explicitly references the use of git diff --staged to retrieve repository changes.

commit