lark-workflow-standup-report
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses imperative language ("CRITICAL", "MUST") to instruct the agent to read a specific file ("../lark-shared/SKILL.md") before starting. This serves as a behavioral override attempting to dictate the agent's operational flow.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources (Lark calendar events and task descriptions) which could contain malicious directives.
- Ingestion points: Content retrieved via 'lark-cli calendar +agenda' and 'lark-cli task +get-my-tasks' in SKILL.md.
- Boundary markers: Absent. The instructions do not specify delimiters to separate untrusted data from system instructions.
- Capability inventory: The skill utilizes shell command execution via the 'lark-cli' tool.
- Sanitization: Absent. There is no instruction to sanitize or escape the event summaries or task descriptions before they are processed by the AI.
- [COMMAND_EXECUTION]: The skill uses the 'lark-cli' binary to perform authenticated network operations to fetch user data.
Audit Metadata