The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/dby.mjs performs dynamic path resolution at runtime. It searches up to four levels of parent directories to locate a core dependency, dby-api/scripts/dby.mjs. This mechanism is used to share a single CLI shim across multiple related skills from the same author.
[COMMAND_EXECUTION]: The skill uses the child_process.spawn method within scripts/dby.mjs to execute its main logic. It passes user-provided text as arguments to a local Node.js process to perform the compliance check.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data (the text to be checked for banned words).
Ingestion points: The user text is ingested via the banned check <文案> command defined in SKILL.md.
Boundary markers: The skill does not implement explicit boundary markers or sanitization for the input text before passing it to the CLI tool.
Capability inventory: The skill has the capability to execute shell commands (via the Node.js shim) and make network requests to the vendor's API (doubaoya.com).
Sanitization: The agent is instructed to parse specific HTML-like tags (<span>) from the API response to identify risks, which relies on the agent's ability to distinguish between data and instructions. The risk is considered low due to the structured nature of the expected output (a markdown table).

dby-banned-words