implement-review
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from code changes (
git diff) and possesses the capability to write to GitHub issues, creating a surface where malicious instructions in the code could attempt to influence agent behavior. - Ingestion points:
git diffoutput processed in Step 1 and Step 2. - Boundary markers: The skill instructions provide a rigid multi-step flow and a specific report template, which helps maintain task focus.
- Capability inventory: Shell command execution via
git,gh issue list,gh issue comment, andcreate-issuecommands. - Sanitization: The skill explicitly instructs the agent to use shell quoting (
"${var}") and heredocs (<<'EOF') to prevent command injection when passing review data to the GitHub CLI. - [COMMAND_EXECUTION]: The skill executes shell commands to analyze the repository state and interact with project management tools.
- Evidence: Commands such as
git diff main...HEAD,gh issue list, andgh issue commentare used to perform the core review and reporting functions. These are gated by instructions requiring user approval for any action that modifies remote state.
Audit Metadata