pnpm-dependency-analysis

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION] (LOW): The skill contains an Indirect Prompt Injection surface where user-provided package names are interpolated directly into shell commands.
  • Ingestion points: The PACKAGE variable used throughout the bash scripts (e.g., in pnpm why "$PACKAGE").
  • Boundary markers: None present. The instructions do not warn the agent to sanitize the input against shell metacharacters.
  • Capability inventory: The skill uses pnpm, grep, find, and jq which can be exploited if the input string contains characters like ;, `, or $(...).
  • Sanitization: None present. External input is used directly in a subshell execution context.
  • [EXTERNAL_DOWNLOADS] (LOW): The skill documentation suggests installing external third-party tools.
  • Evidence: Recommends pnpm install -g pnpm-workspace-graph and brew install ripgrep. While these are useful tools, they represent unmanaged external dependencies that must be manually verified by the user.
  • [COMMAND_EXECUTION] (SAFE): The skill's primary purpose involves executing local CLI tools (pnpm, grep, find) to analyze the repository. These operations are consistent with the stated purpose of dependency analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:17 PM