The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes git diff commands to collect the code changes intended for review. Specifically, it runs git diff @{upstream}...HEAD, git diff main...HEAD, and git diff HEAD~1 to define its working scope.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection due to the way it processes external data.
- Ingestion points: The agent ingests untrusted data from the output of git diff (specified in SKILL.md), which contains code that could be controlled by an attacker or contain malicious comments.
- Boundary markers: Absent. The instructions do not include specific delimiters or warnings to the sub-agents to ignore instructions that might be embedded within the code comments or string literals being reviewed.
- Capability inventory: The skill has the capability to launch four independent sub-agents via the Agent tool and the authority to "apply the fixes" directly to the codebase, which implies file-write capabilities.
- Sanitization: Absent. There is no evidence of sanitization, filtering, or validation of the diff content before it is passed to the analysis agents.

simplify