qqbot-remind
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill interpolates user-controlled data into a message payload that is later processed as instructions by the agent.
- Ingestion point: The {提醒内容} (reminder content) variable in SKILL.md captures user input for reminders.
- Boundary markers: The skill lacks delimiters (e.g., XML tags or triple quotes) or explicit safety instructions to ignore embedded commands within the user-provided content.
- Capability inventory: The skill utilizes the cron tool to schedule future agent executions ('agentTurn'), where the payload.message field containing the user input is interpreted by the model.
- Sanitization: There is no validation or filtering of user input to prevent instruction embedding before it is used in the message template.
Audit Metadata