CoAutoResearchDocumentation

Get started

  • Start your first research project
    • Set up with your coding agent
    • Define your research
    • Choose the agent settings
    • Work together as research progresses
    • Pause, continue or start over
    • Read the manuscript and generate a paper
  • Set up CoAutoResearch with a coding agent
    • 1. Identify the installation
    • 2. Check prerequisites and the selected backend
    • 3. Start and verify the dashboard
    • 4. Prepare paper generation
    • 5. Leave the workspace ready
  • A collaborative research workflow
    • Browse an existing research project
    • Screenshots
    • 1. Define the question and its boundaries
    • 2. Follow the research while discussing it
    • 3. Pause, inspect and continue
    • 4. Check what the research actually established
    • 5. Build and review a paper draft
  • Example papers
  • Research together, build on evidence
    • Give recursive self-improvement a clear target
    • Start with a useful brief
    • Bring existing work without losing its meaning
    • Use discussion to improve the next step
    • Keep each trial focused
      • Resolve execution setup before experiments
    • Make resource searches useful
    • Give guidance at the right boundary
    • Read findings with their limits
    • Move from manuscript to paper in the product
    • Model connection stalls
  • Generate a paper in the dashboard
    • Setup contract for the coding agent
    • Evidence and execution boundaries
  • Frequently asked questions
    • Can I use my existing coding-agent login?
    • Where does my research go?
    • Can I pause and come back?
    • What remains my responsibility?
    • Do I need the npm package?
    • How do I change the documentation theme?
    • Where can I ask a question or report a problem?

Using the system

  • CLI Reference
    • Commands
    • Start the UI
    • Project Dashboard
    • Return to Existing Work
    • Autoresearch Commands
    • v2.0 Run Semantics
    • Update Project Instructions
    • Project Backup And Restore
    • Port and Browser Options
    • Remote Mode
    • Supervised experiment execution
    • Environment Variables
  • Multiple Projects
    • Start A Dashboard
    • Isolation Model
    • Single-Project Mode
  • Remote Servers
    • Cloudflare CLI setup
      • 1. Install cloudflared
      • 2. Run
      • 3. Check
    • Authenticated Production Boundary
    • HTTP proxy servers
    • SSH Fallback
    • File Access Semantics
    • Security
  • Platform Support
    • Required Tools
    • macOS And Linux
    • Windows Native
      • Agent Executable Resolution
      • API-Key Providers
      • Claude Code With Anthropic-Compatible Gateways
    • Windows With WSL2
    • File Attachment Behavior
    • Remote Servers
  • Upgrading
    • Preview And Apply A Project Upgrade
    • Project-data Rollback
    • Legacy Deliverables After Upgrade

Technical reference

  • Concepts
    • Recursive self-improvement
    • System architecture
    • Research Loop
    • What the Diagram Means
    • Critical Path Control
    • Manuscript Deliverables
    • Collaboration boundary
    • Control Surfaces
    • Record validation and repair
    • Compatibility
    • Research ownership
CoAutoResearch
  • Search

Contact: yihong.tang@mail.mcgill.ca

© Copyright 2026, CoAutoResearch contributors.

Built with Sphinx using a theme provided by Read the Docs.