The Agent Skills Directory

[DATA_EXFILTRATION]: The feishu_task_attachment_upload tool supports a file_path parameter, allowing the agent to read local files on the system to upload them as attachments. This could lead to the exposure of sensitive configuration files or credentials if the agent is misled.
[DATA_EXFILTRATION]: The feishu_task_attachment_upload tool also accepts a file_url, which enables the agent to fetch remote content. This functionality can be misused for SSRF (Server-Side Request Forgery) or data exfiltration to external, attacker-controlled servers.
[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection attacks.
Ingestion points: Untrusted content is ingested into the agent context through the feishu_task_get, feishu_task_comment_list, and feishu_tasklist_get tools as defined in the SKILL.md file.
Boundary markers: The skill documentation does not define delimiters or specific instructions to help the model distinguish between system instructions and data retrieved from external sources.
Capability inventory: The skill possesses powerful capabilities including local file reading (feishu_task_attachment_upload), task deletion (feishu_task_delete), and tasklist management (feishu_tasklist_delete).
Sanitization: There is no indication that content retrieved from Feishu is sanitized or validated before being processed by the agent.

feishu-task