Open Research Commons · private by default
Turn a provenance question into evidence others can challenge.
A disciplined workspace for separating what a source says, what it does not say, and what still needs human judgment.
Nothing is uploaded or automatically searched. AI tasks remain proposals; the software cannot authorize identity, novelty, publication, outreach, or payment.
The method
A rigorous packet in four moves
Each move makes the working account easier for another researcher to reproduce, dispute, or correct.
- 01
Read an investigation
Inspect sources, limitations, unresolved questions, and correction history.
Read research leads - 02
Test the method
Open a synthetic packet to examine the schema without mistaking it for evidence.
Open the sample packet - 03
Build privately
Frame a question, record evidence, challenge the account, and keep the draft on your device.
Start the workbench - 04
Return a challenge
Prepare a privacy-minimizing correction or independent-review proposal.
Choose a contribution path
Device-local workbench
Build your review packet
Your entries stay in this browser. The downloaded packet records evidence and uncertainty; it is not a provenance conclusion.
Choose your next step
Use the Commons at the depth you need
Opening a route is an aggregate funnel signal only—not scholarly validation, demand, a sale, revenue, or income.
For review and reuse
Open standards and technical resources
Inspect the method, schemas, software, and machine-readable outputs behind the interface.
- Read the open methodology
- Inspect the production AI evaluation contract
- Machine-readable research feed
- Research software and citation metadata
- Inspect or contribute to the open-source repository
- Open Research Commons JSON-LD catalog
- Research packet JSON Schema
- Bounded AgentTask work-order JSON Schema
- Reproducible synthetic sample packet
- Correction-envelope JSON Schema
- Production-question export JSON Schema
- Qualified-acquisition action JSON Schema