The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection where instructions embedded in smartsheet records could influence agent behavior.
Ingestion points: Data is read from external smartsheets via the wecom-cli smartsheet records query and records list interfaces described in references/smart-sheet-read.md.
Boundary markers: The instructions do not specify the use of clear delimiters or wrapping for data read from sheets before it is processed by the AI.
Capability inventory: The skill possesses extensive write capabilities, including record creation/modification and the ability to send data to external URLs via WeCom webhooks as documented in references/smart-sheet-edit.md and references/smart-sheet-webhook.md.
Sanitization: While SKILL.md contains basic logic to refuse execution if specific prompt injection strings (e.g., '忽略之前的指令') are detected in cell content, this is not a robust sanitization framework.
[DATA_EXFILTRATION]: The skill implements a fallback mechanism in references/smart-sheet-webhook.md that directs the agent to send data to a user-provided Webhook URL. While intended for legitimate data synchronization when CLI limits are reached, this creates a channel for transmitting data to remote endpoints.

wecomcli-smartsheet