The Agent Skills Directory

[SAFE]: The skill is a legitimate tool developed by ByteDance, utilizing its own cloud infrastructure and verified software packages. All network communications are directed to official and trusted ByteDance domains.
[COMMAND_EXECUTION]: The skill executes local commands to manage its runtime environment, including creating a Python virtual environment and running the client script. These operations are standard procedure for skills requiring specific runtime dependencies and isolation.
[EXTERNAL_DOWNLOADS]: During initialization, the skill downloads the official volcengine-python-sdk from a public package registry. This is a trusted dependency required for authenticating with the backend services and is part of the standard setup process.
[CREDENTIALS_UNSAFE]: The skill requires Volcengine API credentials (AccessKey and SecretKey) to function. It includes robust instructions for the agent to handle these securely, requesting them from the user only when needed and strictly prohibiting their exposure or leakage in any output.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it retrieves and displays content from an external API.
Ingestion points: Data from the Volcengine API enters the context via results processed in scripts/openapi_client.py.
Boundary markers: Present; the skill uses fixed markdown templates in SKILL.md for list and detail views to structure the external data.
Capability inventory: Includes shell command execution for environment setup in SKILL.md and network operations in scripts/openapi_client.py.
Sanitization: Absent; API data is placed into formatting templates without explicit escaping or filtering for malicious instructions.

byted-marketing-agent-inspiration-insight