Universal Research Pass v1.0 — Public Alpha Baseline

Universal Research Pass v1.0 — Public Alpha Baseline

This is the first frozen public baseline of the Universal Research Pass.

The Research Pass is an open, community-tested protocol for rigorous AI-assisted research.

This release exists primarily as a stable control condition for community testing.

What v1.0 includes

• Quick Check, Targeted Pass, and Full Research Pass depths
• starting-state freeze
• rival hypotheses
• explanation-space mapping
• vocabulary expansion
• cross-disciplinary search
• source-role labeling
• source-lineage checks
• claim-level evidence appraisal
• triangulation
• counter-evidence and failure searches
• contradiction investigation
• confidence states
• context filtering
• experiment design
• stopping rules
• a simple activation phrase:

Research Pass this: [question, claim, observation, problem, decision, or messy thought]

Why this version is frozen

Community testing is already producing alternative architectures and hypotheses, including:

• modular / skill implementations
• prompt compression
• glossary and definition cleanup
• multi-agent implementations
• cross-vendor testing
• model-specific behavior
• instruction-ablation experiments
• questions about whether modern models need all of the current scaffolding

Those experiments should be compared against an unmoving baseline.

For that reason, canonical v1.0 will remain frozen while alternatives are tested.

A future canonical version must earn its way forward through evidence, comparison, and regression testing.

Project resources

See the repository for:

• RESEARCH_PASS.md
• TESTING_PROTOCOL.md
• TESTER_GUIDE.md
• EXPERIMENTAL_HYPOTHESES.md
• CONTRIBUTING.md
• CHANGELOG.md
• the experimental benchmark suite
• structured failure and improvement issue forms

Current status

Public Alpha

The goal is not to prove that Research Pass works.

The goal is to discover:

where it works, where it fails, and what survives serious attempts to improve or replace it.