alterlab-rma-thesis-architect
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external, untrusted content such as research proposals, drafts, and advisor feedback files to generate structural blueprints and timelines. This creates a surface for indirect prompt injection if the provided files contain malicious instructions.
- Ingestion points: Assessment phase reads project proposals, outlines, drafts, and advisor feedback.
- Boundary markers: Absent. The instructions do not specify delimiters or warnings to ignore instructions embedded within user files.
- Capability inventory: The skill uses search capabilities to research academic standards and write_file capabilities to generate markdown deliverables (blueprints, timelines, prep kits).
- Sanitization: Absent. The skill processes external text to generate summaries and outlines without explicit sanitization steps.
- [SAFE]: The skill's primary function involves generating academic deliverables (blueprints, timelines, chapter outlines) in markdown format. These outputs are restricted to the project directory and use structured naming conventions, which is expected and safe behavior for a writing assistant.
Audit Metadata