duoduo-loop
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
duoduo daemon statuscommand for monitoring and troubleshooting the background scheduler service. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process arbitrary natural language requests through the
/loopcommand to generate and execute background task plans. - Ingestion points: User-provided prose instructions following the
/loopcommand across various interfaces (Feishu, CLI, editors). - Boundary markers: By default, the skill generates an execution plan (detailing cadence, engine, and delivery) which requires user confirmation before task creation. Note that users can explicitly bypass this boundary using a "no need to confirm" waiver.
- Capability inventory: Background tasks have the capability to perform network requests (e.g., monitoring CI, fetching news digests) and perform filesystem operations within the application's runtime directory (
/var/jobs/). - Sanitization: The primary safety mechanism is the human-in-the-loop confirmation flow where the agent echoes the interpreted plan back to the user before execution.
Audit Metadata