riper-5
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow that ingests data from external files during the 'RESEARCH' phase and uses that information to drive code implementation in the 'EXECUTE' phase, creating a vulnerability surface for indirect prompt injection.
- Ingestion points: The
RESEARCHmode explicitly permits the agent to '阅读文件' (read files) and '理解代码结构' (understand code structure) from the project environment. - Boundary markers: The protocol lacks instructions for the agent to treat file content as untrusted or to use specific delimiters to separate embedded data from instructions.
- Capability inventory: The skill facilitates file reading, task file creation, and code implementation ('实施') via the 'EXECUTE' mode.
- Sanitization: There are no requirements for the agent to sanitize, filter, or validate the content of read files before using them to generate technical specifications or implementation steps.
Audit Metadata