The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the lark-cli binary to perform all operations, executing commands such as lark-cli skills list, lark-cli skills read, and domain-specific commands for document management and authentication.
[DYNAMIC_EXECUTION]: The skill dynamically loads instructions from version-matched guides using lark-cli skills read and executes the commands defined within those guides. The instructions include a requirement to verify file integrity using SHA-256 when falling back to disk-based guides, which serves as a security control against tampering.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources including Lark/Feishu documents, wiki pages, and IM messages through the Lark CLI, which may contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Document content, wiki space nodes, and IM message threads retrieved via the CLI.
Boundary markers: Instructions to treat metadata.cliHelp as a recipe rather than a literal command and to verify business success by inspecting JSON response fields (ok, errors) rather than relying solely on process exit codes.
Capability inventory: Shell command execution via the lark-cli binary.
Sanitization: Instructions emphasize following domain-specific verification contracts and performing independent reads to confirm state changes after write operations.

lark-cli-router