agent-code-review

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from GitHub Pull Requests, including code changes and PR descriptions, which could contain malicious instructions designed to influence the review agent's behavior.
  • Ingestion points: Pull Request metadata and file contents are fetched from GitHub repositories using the gh CLI and the agent-code-review tool as documented in SKILL.md.
  • Boundary markers: The provided instructions do not specify explicit delimiters or warnings to separate untrusted PR content from the agent's internal instructions.
  • Capability inventory: According to SKILL.md, the agent can execute shell commands via the agent-code-review binary and has access to file system tools like Read, Grep, and Glob.
  • Sanitization: No specific sanitization or validation logic for the ingested PR content is described in the skill metadata.
  • [COMMAND_EXECUTION]: The skill makes extensive use of the agent-code-review CLI tool to manage the review queue, author roster, and server configuration. It also relies on the availability of gh, duckdb, and AI inference engines (codex or claude) on the system PATH.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 11:51 PM
Security Audit — agent-trust-hub — agent-code-review