js-best-practices
Pass
Audited by Gen Agent Trust Hub on Nov 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a provided Node.js script (
scripts/js-profile.mjs) to automatically detect project-specific settings such as package managers and compiler options. The script is part of the skill's own codebase and uses standard Node.js modules to inspect the filesystem locally. - [INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface where instructions could potentially be ingested from untrusted project files.
- Ingestion points: Project configuration files like
package.json,tsconfig.json, and lockfiles are read by thejs-profile.mjsscript. - Boundary markers: The profiling output is a generated markdown file that does not use specific delimiters to distinguish between project data and agent instructions.
- Capability inventory: The skill allows the agent to read and write files within the project's
.agents/directory to manage context-aware profiles. - Sanitization: Data from project configuration files is incorporated into the profile without explicit sanitization or validation to prevent command or instruction injection.
Audit Metadata