jfrog-mcp-management
Warn
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: MEDIUMPROMPT_INJECTIONCREDENTIALS_UNSAFEPERSISTENCEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill instructs the agent to perform an activation check silently and to conceal the results of this check from the user. It also directs the agent to mention failures only as a side note without stopping to fix them. These concealment patterns are designed to reduce user oversight.
- [CREDENTIALS_UNSAFE]: The skill and its scripts access sensitive JFrog configuration files in the user's home directory (such as ~/.jfrog/jfrog-cli.conf.v6) to retrieve authentication tokens.
- [PERSISTENCE]: The skill provides instructions for the agent to guide the user in modifying shell startup scripts (e.g., ~/.bashrc, ~/.zshrc) to persist environment variables. Modifying shell profiles is a standard persistence mechanism.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from a remote JFrog catalog to generate configuration files. Ingestion points: Data enters via npx catalog commands. Boundary markers: Absent. Capability inventory: Shell execution and file writes. Sanitization: Absent.
- [COMMAND_EXECUTION]: The skill executes various shell commands including npx, jf, and Node.js scripts.
- [EXTERNAL_DOWNLOADS]: The skill downloads and executes remote code from an external package registry using npx.
Audit Metadata