full-review

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides a structured framework for code review, including comprehensive reference guides for security patterns, common bugs, performance, and architecture. It follows best practices for static analysis and risk assessment by defining clear confidence levels and "Do Not Flag" lists to avoid false positives.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data through the review of local source code files. Ingestion points: The agent reads source files from the local repository using Read, Glob, Grep, and Bash tools (SKILL.md). Boundary markers: The skill instructions do not explicitly implement markers to isolate code content from agent instructions, though Step 4 (Verify & Report) encourages cross-referencing findings. Capability inventory: The agent has access to Bash, Write, Read, and Agent tools, enabling file modification and shell command execution. Sanitization: No explicit sanitization or filtering of the source code content is performed before processing. This vulnerability surface is inherent to the skill's primary function.
  • [COMMAND_EXECUTION]: The skill uses standard local commands like git diff and find to establish context and scope. These operations are restricted to local directory enumeration and diffing for the purpose of identifying changed files, posing no risk of unauthorized data access or network exfiltration.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 12:08 PM