bundle-analysis

Warn

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands such as 'npm run build', 'yarn build', and 'pnpm build' to generate build statistics. It also uses shell redirection to create 'stats.json'.\n- [DATA_EXFILTRATION]: The skill attempts to access and interact with the directory path '~/.claude/projects/', which is outside the current project scope. Accessing or writing to platform-specific state directories in the user's home folder can lead to the exposure of sensitive platform metadata or the persistence of data across sessions.\n- [PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection because it processes untrusted data from the local environment and project files.\n
  • Ingestion points: Build configuration files (e.g., webpack.config.js, vite.config.ts) and dependency lock files (package-lock.json, yarn.lock).\n
  • Boundary markers: None detected; instructions do not provide delimiters to separate untrusted data from agent instructions.\n
  • Capability inventory: High-privilege capabilities including shell command execution and file system writes.\n
  • Sanitization: No evidence of validation or sanitization of the content extracted from external project files.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 23, 2026, 10:56 AM
Security Audit — agent-trust-hub — bundle-analysis