The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests untrusted data from external Feishu (Lark) documents, creating a surface for Indirect Prompt Injection (Category 8).
Ingestion points: Content fetched from user-provided URLs using the lark-cli docs +fetch --doc command in the '了解需求' (Understand Requirements) step.
Boundary markers: Absent. The skill instructions do not define delimiters or provide specific instructions for the agent to ignore potentially malicious commands embedded within the fetched document content.
Capability inventory: Subprocess execution via lark-cli, file system write access (writing to /docs/prd-sumerize.md and /prototype/), and the ability to invoke other skills like /figma-prototype.
Sanitization: Absent. The fetched content is summarized and integrated directly into the agent's context and technical solution documents.
[COMMAND_EXECUTION]: The skill relies on executing external CLI tools and other skills to fulfill its logic.
Evidence: Execution of the lark-cli docs +fetch --doc command which interacts with the network to retrieve document data.
Evidence: Calls to the /figma-prototype skill which generates local files and prototype assets based on user-provided Figma links.

requirements