volcengine-cli

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly asks for AccessKey/SecretKey and authorization codes and instructs the agent to pass them verbatim into shell commands (e.g., ve configure set --access-key/--secret-key and scripts/ve_login_remote.sh complete ), forcing the LLM to handle/output secrets directly.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). At runtime, the skill’s scripts/fetch_swagger.py (used via python3 scripts/fetch_swagger.py ...) fetches API documentation text from Volcengine’s OpenAPI Explorer over the network and converts it into markdown that the agent ingests, with no requirement that the fetched content be first-party/trusted in the threat model.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill includes runtime code that fetches API definitions/search results from https://api.volcengine.com/api/common/explorer and https://api.volcengine.com/api/common/search/all?{params}, which are retrieved at runtime (scripts/fetch_swagger.py and scripts/find_api.py) and used to generate documentation/lookup data that directly influence the agent's prompts/command construction.