The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the @larksuite/cli package from the NPM registry during setup if it is missing from the environment. This is the official command-line tool for the Lark/Feishu platform.
[COMMAND_EXECUTION]: The lark_setup.py script executes shell commands to open a browser for user authentication and to initialize the CLI configuration. It uses subprocess.run with list-based arguments and passes secrets via standard input (stdin) to avoid exposing credentials in the process list or command history.
[PROMPT_INJECTION]: The skill exposes a potential attack surface for indirect prompt injection as it is designed to process external data.
Ingestion points: Untrusted data enters the agent context through emails (lark-mail.md), instant messages (lark-im.md), and cloud documents (lark-doc.md).
Boundary markers: No explicit delimiter-based boundary markers are implemented in the code, but the instructions provide clear warnings to the agent.
Capability inventory: The skill has access to the Bash tool and extensive platform APIs for reading and writing data, which could be leveraged if an injection succeeds.
Sanitization: The skill documentation (lark-mail.md) explicitly warns the agent to treat email content as untrusted input and apply sanitization and validation.

lark-unified