The Agent Skills Directory

[SAFE]: The skill performs repository analysis by cloning remote Git repositories or accessing local paths to identify frameworks and dependencies. This operation is limited to diagnostic pattern matching using grep and find, with no execution of the target source code.
[SAFE]: The skill integrates with the official Volcengine CLI ('ve') for identity verification and region probing. All cloud operations are performed using standard vendor-provided tools and environment variables.
[SAFE]: Secure practices are observed, such as explicitly instructing the agent to avoid accessing sensitive local configuration files (~/.volcengine/config.json).
[INDIRECT_PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection due to its processing of external repository content. \n
Ingestion points: Clones remote Git repositories or reads local files as specified in SKILL.md. \n
Boundary markers: None are defined to differentiate between repository data and agent instructions. \n
Capability inventory: The skill can execute shell commands, perform network calls via git and ve, and manage local state in the .volcengine/ directory. \n
Sanitization: There is no evidence of content sanitization or filtering for the source code files analyzed. This risk is managed by the skill's read-only analysis context.

volcengine-prepare