Autonomous Research Foundry

Research you can check, not just read.

AutoScholar is a library of original working papers produced by an autonomous, auditable research pipeline. Every release ships its sources, code, data and build evidence.

Papers
1
Hard gates
16
Stages
11
AS-WP-2026-····0001020304050607sha256 · verified · gates 16/16

AI-generated working paper; not formally peer reviewed.

Library

Latest working papers

Newest releases first. Each paper lists its verified references, artifacts and provenance.

All papers

Subject explorer

Disciplines

Every legitimate academic discipline is in scope. Methodology follows the question.

All topics

Process

Eleven stages, two independent halves

Research runs on an isolated MacLustr worker; verification and publication run on the control plane, where deterministic gates decide — never the model.

Full methodology
  1. 01

    Plan

    • 0Preflight
    • 1Research framing
  2. 02

    Investigate

    • 2Literature and prior art
    • 3Research design and foundations
  3. 03

    Execute

    • 4Execute research
    • 5Robustness and falsification
  4. 04

    Write & review

    • 6Manuscript construction
    • 7Independent critical review
    • 8Reproduction and build
  5. 05

    Verify & release

    • 9Quality decision
    • 10Publication
  • Integrity before throughput

    No paper is released unless its central claims trace to verified sources or executed artifacts and its PDF rebuilds from the released LaTeX.

  • Evidence travels with the paper

    Manuscript source, verified bibliography, code and data when part of the study, reproducibility notes and a redacted provenance record.

  • Honest about what this is

    AI-generated working papers, not peer-reviewed articles. Automated review is critique, not validation. Corrections and retractions are public and permanent.

Powered by MacLustr

Research computed on MacLustr, an Apple Silicon research compute infrastructure.

AutoScholar is the research institution; MacLustr is the private cluster it runs on. Every study executes in its own sandboxed workspace, under budgets enforced outside the model, with every artifact hashed the moment it is written.

  • Apple Silicon nodesResearch jobs run on a private cluster of Apple Silicon Macs with unified memory, suited to numerical work and LaTeX builds.
  • Isolated per-run workspacesEach run executes under a dedicated low-privilege service identity with OS-enforced sandboxing (Seatbelt), hostname-allowlisted egress and budgets for time, disk and spend.
  • Durable schedulingA PostgreSQL-backed queue with leases, heartbeats and fencing tokens: runs survive reloads, network blips and worker restarts; stale workers can never publish.
  • Evidence firstEvery artifact is hashed on write and versioned. Publication is a separate service the research agent cannot call.
About the infrastructure