A method for testing what least-privilege access would remove, break, and protect, before redesigning every agent.
What the research pack covers
The asset walks a research lead through harvesting assumptions, grading them, ranking materiality, the pilot, task failures, just-in-time access, revocation, attribution, blast radius, and freezing the design. It ends with an assumption register, a source log, and research questions.
- Turning a baseline into governed access assumptions
- Piloting least privilege for ten agents
- Measuring task failures from missing access