The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill manages the lifecycle of the @volcengine/vefaas-cli Node.js package. This tool is the official CLI for Volcengine's cloud platform and is used for all core functionalities including deployment and resource management.
[INDIRECT_PROMPT_INJECTION]: The skill exposes a data ingestion surface area that could theoretically be used for indirect prompt injection.
Ingestion points: The agent is instructed to fetch and process external documentation using vefaas doc and to read runtime output and system logs using vefaas logs and vefaas sandbox logs (File: references/vefaas-docs.md, references/vefaas-function.md).
Boundary markers: The skill frequently mandates the use of structured data formats (e.g., --output json) and specific field extraction (e.g., data.markdown) to limit the influence of unstructured external content.
Capability inventory: The agent has extensive capabilities, including creating/deleting cloud functions, modifying production environment variables, and executing arbitrary OpenAPI actions (File: references/vefaas-openapi.md, references/vefaas-application.md).
Sanitization: The instructions focus on structured parsing but do not explicitly detail sanitization routines for the content of logs or third-party code processed during inspect or deploy tasks.
[CREDENTIALS_SAFE]: The skill contains explicit security guardrails preventing the exfiltration of sensitive information. It instructs the agent to never echo Access Keys (AK), Secret Keys (SK), session tokens, or .env values back to the user (File: SKILL.md, references/vefaas-auth.md).

vefaas