cursor-ralph-loop
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for iterative, self-referential development that processes user-controlled inputs across multiple steps, creating an attack surface for instructions embedded in the task data.
- Ingestion points: User-provided task descriptions and project scope records defined in the
SKILL.mdworkflow. - Boundary markers: The instructions do not establish clear delimiters or use "ignore instructions" wrappers for the external task data.
- Capability inventory: The skill utilizes the agent's native iteration and worker identity management capabilities as described in
SKILL.md. - Sanitization: There is no specified logic for sanitizing or filtering the content of the tasks being iterated upon.
- [EXTERNAL_DOWNLOADS]: The
MIRROR.mdfile contains references to the official source repository and documentation for the skill hosted on GitHub.
Audit Metadata