The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external Lark/Feishu documents, which represents a potential attack surface.
Ingestion points: External content is loaded into the agent's context through document reading commands like docs +fetch and mindnotes nodes list (refer to lark-doc-fetch.md and lark-doc-mindnote.md).
Boundary markers: The skill employs structured XML and Markdown tagging to delineate and organize document content.
Capability inventory: The agent can execute a variety of document and file system operations via the lark-cli tool, including creating documents and downloading resources.
Sanitization: The instructions specifically mandate that the agent must validate URL hosts before fetching content, requiring it to reject private, loopback, and local IP addresses to prevent Server-Side Request Forgery (SSRF) and unauthorized internal access (detailed in lark-doc-fetch.md and lark-doc-resource-cover.md).
[COMMAND_EXECUTION]: The skill uses the lark-cli command-line interface to execute all document management tasks. This usage is transparently documented in the reference files as the primary and expected method of operation for this skill.
[SAFE]: The skill demonstrates high security maturity by incorporating defensive instructions (such as URL filtering and relative path restrictions) into its operational workflows to mitigate common risks associated with processing external content.

lark-doc