modern-javascript
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions create a potential attack surface for indirect prompt injection by directing the agent to ingest and act upon data from external, untrusted project files.
- Ingestion points: The agent is instructed to "Inspect
engines, browser targets, lockfiles, and build configuration as needed" (SKILL.md) and to evaluate "actual targets:engines,browserslist, deployment images... and bundler/test configuration" (references/COMPATIBILITY.md). - Boundary markers: The skill lacks requirements for the agent to use specific delimiters or instructions to disregard embedded commands/injections when parsing these external files.
- Capability inventory: As a developer-oriented skill, the agent is expected to have capabilities to write files and potentially execute commands, which increases the impact of successful injection from malicious project metadata.
- Sanitization: No sanitization or validation protocols are established for the data ingested from project configuration files before it is processed by the agent.
Audit Metadata