The Agent Skills Directory

[SAFE]: The skill implements a stylistic code review process using specific personas. No malicious instructions or behavior overrides were detected.
[SAFE]: No network operations, credential harvesting, or sensitive file access patterns were identified. The skill's operations are confined to the local development environment and standard code review inputs.
[SAFE]: The logic regarding parallel analysis tasks is a platform-specific optimization request for environments like Copilot and does not represent a privilege escalation or security risk.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data (code from files or git diffs) without explicit boundary markers or sanitization logic.
Ingestion points: Reads code from paths provided in $ARGUMENTS or from git diff output.
Boundary markers: Absent; the skill does not instruct the agent to use specific delimiters or ignore instructions found within the code comments of the reviewed files.
Capability inventory: Limited to generating text-based review conclusions and structured tables.
Sanitization: Absent; the skill relies on the agent's base safety layers to handle potential injections within the code it reviews.

review-gal