deobfuscate-javascript
Fail
Audited by Gen Agent Trust Hub on Aug 9, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The script 'scripts/unpack.ts' uses 'new Function()' to evaluate segments of the input JavaScript code to reverse-engineer obfuscation techniques like 'Packed' or 'AAEncoded' formats. If the input file provided by a user is maliciously crafted, this evaluation can result in arbitrary code execution on the host machine.
- [EXTERNAL_DOWNLOADS]: Scripts such as 'wakaru-normalize.ts' and 'format.ts' use 'npx' or 'bunx' to fetch and run external packages like '@wakaru/cli' and 'prettier' from the npm registry at runtime, which introduces supply chain risks.
- [PROMPT_INJECTION]: The 'Acceptance Review' process (described in 'scripts/acceptance-checklist.md') requires the AI agent to read deobfuscated code end-to-end to assess quality. This untrusted content serves as a vector for indirect prompt injection, where malicious instructions hidden in the analyzed code could hijack the agent's behavior.
- [COMMAND_EXECUTION]: The skill uses 'spawnSync' across various scripts to orchestrate the execution of internal and external CLI tools, increasing the attack surface if the agent is manipulated into running malicious commands.
Recommendations
- AI detected serious security threats
Audit Metadata