jd-aipay-onboarding
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill manages dependencies for its server-side examples, including standard libraries such as sm-crypto (Node.js), gmssl (Python), and cryptography (Python). It also includes a bundled JAR file (wyaks-security) for the Java demo, which is a vendor-provided security component. The skill itself is designed to be installed via npx from the @jdpay/aipay npm package.
- [COMMAND_EXECUTION]: The provided CLI tool (bin/cli.js) performs file system operations (creating directories, copying files) to install the skill content into local agent directories (~/.claude/skills, ~/.cursor/rules, etc.). This is the intended administrative function of the setup script.
- [OBFUSCATION]: Deterministic detectors identified large hex blobs in the cryptographic utility files (pfx.js, pfx.py). Analysis confirms these are benign implementation constants for the RC2 block cipher, which is necessary for parsing specific PKCS#12 certificate structures used in the JD payment ecosystem.
- [INDIRECT_PROMPT_INJECTION]: The skill instructions allow the agent to read and modify user server-side code. This creates a surface for indirect injection if malicious instructions are present in the user's project files. However, the skill provides specific playbooks to manage security, such as ensuring that sensitive credentials are not exposed in responses or logs.
Audit Metadata