Research library

The complete record, including the misses.

Studies enter this registry by identifier and protocol hash. A blocked study remains blocked until its evidence and disclosure gates pass.

IDTitleClassStatusOutcome
HNI-2026-001Tool Scope, Rules Context, and Token Economics in a Local Hermes Agent Harnessexploratoryevidence checkedsupports
Experiment Note

Bounded exploratory evidence.

Useful for mechanism discovery, with limitations stated before any general claim.

Registered Study

Protocol fixed before data.

Hypotheses, outcomes, sample, exclusions, stopping, and analysis are locked first.

Negative Result

A finding worth keeping.

Failed interventions and unresolved hypotheses prevent the Institute from becoming its own confirmation engine.