The Agent Skills Directory

[SAFE]: The skill uses a vendor-specific binary wecom-cli for all operations and follows strict, multi-step workflows for input validation and participant resolution.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from WeCom schedule subjects and descriptions. This presents an indirect prompt injection surface if an attacker-controlled schedule contains malicious instructions intended to mislead the agent. 1. Ingestion points: schedule metadata retrieved via list, search, and get operations. 2. Boundary markers: None defined. 3. Capability inventory: Write/Delete access to schedules and coordination of meeting rooms via wecom-cli. 4. Sanitization: None mentioned for retrieved text fields.

wecomcli-calendar