baoyu-image-gen
Warn
Audited by Gen Agent Trust Hub on Mar 9, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/providers/google.tsusesexecSyncto runcurlfor API requests when an HTTP proxy is detected. It constructs the command using string interpolation of environment variables. - Evidence: In
postGoogleJsonViaCurl, the code buildsproxyArgsusingprocess.env.HTTPS_PROXY(or similar) and interpolates it directly into the shell command string:curl ... ${proxyArgs} "${url}" .... - Risk: This pattern is vulnerable to command injection if an environment variable is set to a malicious value containing shell metacharacters (e.g., backticks or semicolons).
- [DATA_EXFILTRATION]: The skill is configured to read sensitive environment files and transmit data to external services.
- Evidence: The script reads API keys and configuration from
~/.baoyu-skills/.envand project-local.envfiles. It also reads user-specified local files via--promptfilesand--refarguments to send their content to external AI providers (OpenAI, Google, Alibaba, Replicate). - Context: While this is core to the skill's functionality, it establishes a data flow from the local filesystem to external network endpoints.
Audit Metadata