resolve-human-reviews
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches and executes the
agent-reviewspackage from the public npm registry using package runners such asnpx,pnpm dlx,yarn dlx, orbunx. - [REMOTE_CODE_EXECUTION]: Code from the external
agent-reviewspackage is executed in the agent's environment to facilitate PR interaction and code evaluation. - [COMMAND_EXECUTION]: The instructions direct the agent to execute multiple shell commands, including
gitfor repository management (config,add,commit,push) and various Node.js package execution commands. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill ingests untrusted data in the form of human review comments fetched via the CLI (
SKILL.md). - Boundary markers: There are no boundary markers or sanitization steps defined to prevent the agent from following malicious instructions that might be embedded in a PR comment.
- Capability inventory: The agent possesses significant capabilities, including reading and modifying the codebase, executing project scripts (lint/type-check), and pushing changes to remote repositories.
- Sanitization: The skill lacks logic to sanitize or escape the content of the review feedback before the agent processes and acts upon it.
Audit Metadata