oneos-pm-product
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it is designed to ingest and process untrusted user input ("Who does what") to generate functional specifications and prototypes.
- Ingestion points: User-provided requirements, tasks, and stories processed via the "instruction compiler."
- Boundary markers: The instructions mandate reading sibling files (
../yanchufasui/boot.mdand../yanchufasui/habits.md) which serve as runtime protocols to govern agent behavior and handling of input. - Capability inventory: Writing specifications to local files (
.spec/requirements-prd.md), updating CSS themes (oneos-v2-theme.css), and interacting with project management protocols (warRoomSyncProtocol). - Sanitization: The skill delegates validation to a self-check process ("habits §3") and human-in-the-loop triggers ("Y2b 一问") to resolve ambiguities before action.
Audit Metadata