State before story
The project favors claims that can be connected to inspectable runtime state.
Methods
Evidence philosophy.
Cognitive Runtime Project is built around a simple public standard: claims about AI cognition should be tied to visible state, repeatable observations, and clear limits.
Behavior that can be revisited
Evidence should leave behind enough structure for later review, comparison, and replication.
Claims with clear scope
The site separates what has been shown, what is being tested, and what remains open.
Public Methods
Enough method to trust the work, without crowding the story.
Public method notes stay compact and focus on the evidence readers need to evaluate the work.
Public
Readable summaries
Methods should explain what was tested in language suitable for external readers.
Artifacts
Details live with papers and supplements
Deeper data and analysis material can be linked from paper-specific releases.