The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes untrusted source code and git diffs, creating a surface where malicious instructions embedded in a pull request could influence the AI agent's evaluation. * Ingestion points: scripts/pr_analyzer.py and scripts/code_quality_checker.py read files and git data from the target repository. * Boundary markers: Absent; analysis findings are output as structured text or markdown without markers to separate untrusted content from tool metadata. * Capability inventory: The skill uses subprocess.run to execute git commands and local analysis scripts. * Sanitization: Absent; the tool does not filter or escape analyzed code snippets to prevent agent manipulation.
[COMMAND_EXECUTION]: The skill interacts with the local system by executing Git CLI commands and internal analysis scripts through subprocess.run. These calls are constructed safely using list-based arguments without shell=True, minimizing risks from malformed repository paths or branch names.

code-reviewer