The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes wecom-cli commands to list sessions and send messages. These are legitimate operations within the scope of the tool's purpose.
[PROMPT_INJECTION]: The skill manages indirect prompt injection risks associated with processing external data. Ingestion points include the sessions list API output and user-provided message content. While the instructions do not define textual boundary markers, they establish a secure process where the agent must verify and copy target session IDs from a freshly retrieved API response. The capability inventory includes sending messages via wecom-cli. Sanitization is enforced by prohibiting the use of user-supplied or cached identifiers, ensuring the agent only interacts with validated sessions.

wecomcli-message