The Agent Skills Directory

[SAFE]: The skill provides comprehensive and standard documentation for the lark-cli binary. All identified shortcuts and CLI methods are aligned with the stated purpose of calendar management. No obfuscation, persistence mechanisms, or unauthorized privilege escalation patterns were found.- [INDIRECT_PROMPT_INJECTION]: The skill defines a potential attack surface by allowing user-provided content to be interpolated into the --description flag of lark-cli commands. The underlying tool supports a file-read prefix (@) which could be abused to read local files if an attacker successfully injects instructions into event metadata. This is considered a low-risk capability inherent to the productivity tool's design.
Ingestion points: Event title and description fields processed in SKILL.md and references/lark-calendar-create.md.
Boundary markers: The instructions mandate a human-in-the-loop confirmation step for all write operations and require explicit mapping of event IDs, which mitigates automated exploitation.
Capability inventory: Execution of the lark-cli binary with the ability to read files from the working directory and perform network uploads to the vendor's cloud service.
Sanitization: The skill provides instructions for Markdown formatting and time conversion but relies on user confirmation rather than specific input sanitization to prevent the misuse of file-read prefixes.

lark-calendar