reverse-engineer-js
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill utilizes
npxto download and execute tools from the NPM registry, specificallyhumanifyjs,webcrack, andprettier. This is a standard developer practice but involves fetching code from a public registry at runtime. - [COMMAND_EXECUTION]: The skill provides instructions for executing various shell commands to manage files, archive bundles, and invoke deobfuscation scripts. It also uses the
scriptutility to emulate a TTY environment, which is a common requirement for tools with progress bars running in agent contexts. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted, user-provided JavaScript bundles through an LLM-assisted tool (
humanifyjs). This creates an attack surface where obfuscated code could potentially be crafted to influence the LLM's output or the agent's interpretation of the logic. - Ingestion points: Untrusted JavaScript files provided by the user (e.g.,
bundle.js,inputFile). - Boundary markers: No specific boundary markers or "ignore" instructions are added to the code before processing.
- Capability inventory: The skill enables subprocess execution (
npx), file system operations (cp,mkdir), and script execution (node). - Sanitization: No sanitization or validation is performed on the input JavaScript before it is sent to the LLM for renaming.
Audit Metadata