comprehensive-review-full-review
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external, untrusted code provided through the
$ARGUMENTSvariable. This content is then distributed across multiple specialized sub-agents (e.g.,security-auditor,architect-review,code-reviewer) for deep analysis. This workflow presents a vulnerability surface for indirect prompt injection, where an attacker could embed malicious natural language instructions within the code (such as in comments or strings) to influence or bypass the sub-agents' analysis logic. - Ingestion points: The skill receives untrusted data via the
$ARGUMENTSplaceholder inSKILL.md, which is used in prompts for all four review phases. - Boundary markers: There are no explicit delimiters or instructions provided to the sub-agents to treat the input code as data only or to ignore any embedded instructions.
- Capability inventory: The skill utilizes a
Tasktool to orchestrate sub-agents capable of performing static analysis, security vulnerability assessments (using tools like Snyk, Trivy, and GitLeaks), and architectural reviews. - Sanitization: The skill does not define any sanitization, filtering, or validation steps for the input code before it is passed to the sub-agents.
Audit Metadata