The Agent Skills Directory

[DATA_EXPOSURE]: The skill analyzes local files and repository contents to identify frameworks and dependencies. This is a core part of its functionality. It explicitly instructs the agent not to read the user's local ~/.volcengine/config.json file to prevent accidental exposure of stored secrets, demonstrating a security-conscious design.
[EXTERNAL_DOWNLOADS]: The skill clones remote Git repositories to a temporary directory for analysis when a Git URL is provided. This behavior is expected and primary to the skill's purpose. The use of a regular expression to validate the input URL (^(https?|git@)) provides a basic layer of protection against command injection.
[COMMAND_EXECUTION]: The skill executes shell scripts and the official Volcengine CLI (ve) to perform project analysis and cloud service availability checks. These operations are consistent with the vendor's stated purpose and use standard system utilities (find, grep, jq).
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the analyzed repositories. While this presents a theoretical attack surface, the analysis is performed by a script that outputs structured JSON, and the agent uses jq to process the results, minimizing the risk of the agent interpreting data as instructions. This is considered a low-risk profile.

volcengine-prepare