What problem does it solve?
This Skill prevents long-running AI-assisted research from drifting through unsupported assumptions, undocumented decisions, repeated failed experiments, and untraceable implementation changes.
Core Features & Use Cases
- Evidence-Driven Workflow: Connect observations, hypotheses, experiments, decisions, and conclusions to authoritative sources and explicit evidence tags.
- Durable Research Memory: Organize project metadata, knowledge state, experiment logs, question queues, implementation plans, and source registers for continuity across AI sessions.
- Validation and Safety Gates: Require source tracing, falsifiable hypotheses, named observables, regression checks, source archaeology, and goal stops before behavior-changing modifications.
- Focused Research Profiles: Apply specialized safeguards for paper reproduction and numerical-solver diagnosis, including paper tracing, figure calibration, numerical isolation, and reconstruction controls.
- Use Case: Apply the Skill when reproducing a scientific paper, diagnosing solver convergence, validating a specification, or continuing a research project across multiple AI sessions without losing evidence or decision history.
Quick Start
Use the ai-research-protocol skill to map the project's authoritative sources, current knowledge, open questions, experiment history, and next highest-value research action before making any behavior-changing edit.
