guided-code-review

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands such as git log and git show to inspect commit history and changes. It also suggests using network-enabled CLI tools like httpie or wget to fetch project-related artifacts such as OpenAPI specifications or database migrations to build a mental model of the changes during the context-gathering phase.\n- [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it processes data from external, untrusted sources including commit messages, source code files, and ticket descriptions. Without sanitization or boundary markers, malicious instructions embedded in these sources could potentially influence the agent's behavior during the review.\n
  • Ingestion points: Commit messages and code diffs retrieved via git, as well as story or ticket descriptions provided by the user or external systems in Phase 1.\n
  • Boundary markers: Absent. The instructions do not specify delimiters to isolate ingested data from the agent's operational logic.\n
  • Capability inventory: The skill uses shell command execution (git, wget, httpie), local file system access, and the question/questionnaire tool.\n
  • Sanitization: Absent. No validation or filtering is applied to the content of commits or ticket descriptions before the agent processes them.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 07:14 AM