DraftPilot

The local-first AI workspace for research papers

Write and manage LaTeX papers with a local Overleaf-like experience while staying fully connected to your codebase, figures, PDFs, experiments, reviews, rebuttal notes, and AI agents.

DraftPilot combines manuscript editing, PDF compilation, SyncTeX navigation, highlights, comments, TODOs, and AI-assisted research workflows into one unified workspace built for modern researchers.

Download CLI Read Docs Join Discord

Research writing shouldn't be disconnected from research itself.

Most paper-writing tools live separately from your code, experiments, figures, datasets, and AI workflows.

code experiments figures datasets AI workflows

DraftPilot bridges that gap. Your manuscript, implementation, PDFs, figures, logs, and repositories stay connected in one local-first environment designed for AI-native research workflows.

How It Works

Keep the project in GitHub. Run DraftPilot wherever the files live.

Use a private GitHub repo for the paper folder, clone it on your laptop or VM, then launch DraftPilot against that real folder. DraftPilot edits local files, compiles locally, and gives Codex or another agent the same project context you see.

Private GitHub repopaper, figures, bib, notes

→

Laptop or VM clonereal filesystem paths

→

DraftPilot launchLaTeX, PDF, logs, comments, agent

Typical repo workflow

git clone git@github.com:you/paper.git
cd paper
draftpilot launch --paper . --main-tex main.tex --codebase ../research-code --open

Pick Your Workflow

One product, three common research setups.

DraftPilot is designed for sprinting between machines without changing how your paper is stored. The paper can live in a private repo, and the app runs next to the clone you are editing today.

Local

Paper and code on your laptop

Clone the paper repo, attach your local codebase, and open the browser workspace for writing, compiling, comments, highlights, and agent help.

draftpilot launch \
  --paper ~/papers/my-paper \
  --main-tex main.tex \
  --codebase ~/code/my-project \
  --open

Experiments live on a GPU VM

Install DraftPilot on the VM, launch beside the remote paper and codebase, and use the local tunnel link from your laptop.

draftpilot launch \
  --host user@vm.example.com \
  --paper /home/user/paper \
  --main-tex main.tex \
  --codebase /home/user/code \
  --open

Multi-device

Move between devices without losing context

Commit the paper folder and DraftPilot metadata to a private repo when appropriate, clone it elsewhere, then continue from the same TeX files, figures, TODOs, and review notes.

git pull
draftpilot launch --paper . --main-tex main.tex --open

Git-Native Research

Use GitHub for storage. Use DraftPilot for the workspace.

DraftPilot does not replace GitHub. GitHub is where your paper repo, figures, bibliography, and source history can live. DraftPilot is the local workspace that opens that repo, compiles the PDF, stores review metadata, and connects the manuscript to the codebase used to produce the results.

Paper repoTeX files, figures, bibliography, notes, and optional DraftPilot metadata for comments, TODOs, highlights, linked codebases, and deadlines.

Local build outputGenerated PDFs, SyncTeX, aux files, and compile logs stay on the machine running DraftPilot unless you choose to commit or share them.

Codebase linkDraftPilot records which code repositories belong to the paper so Codex can reason across claims, experiments, figures, and implementation details.

VM/device sprintingClone or pull the repo where you want to work, run DraftPilot there, and keep writing with the local files available on that machine.

Download

Install one binary. Launch the workspace where your research lives.

DraftPilot ships as a standalone CLI for macOS, Linux, and Windows. The source repo stays private; users install the binary, point it at a paper folder and a codebase, and open the local browser link.

macOS Desktop App

The desktop app is in preview packaging. Use the macOS CLI binary below for the current public release.

Desktop app status

Current public release:
draftpilot-0.1.3-macos-arm64.tar.gz

Linux Desktop App

The AppImage build is not published yet. Use the Linux CLI binary below for laptops, workstations, and VMs.

Desktop app status

Current public release:
draftpilot-0.1.3-linux-x64.tar.gz

macOS Apple Silicon

For M-series Macs.

Download macOS arm64

tar -xzf draftpilot-0.1.3-macos-arm64.tar.gz
chmod +x draftpilot-macos-arm64
sudo mv draftpilot-macos-arm64 /usr/local/bin/draftpilot
draftpilot help

Linux x64

For Linux laptops, workstations, and VMs.

Download Linux x64

tar -xzf draftpilot-0.1.3-linux-x64.tar.gz
chmod +x draftpilot-linux-x64
sudo mv draftpilot-linux-x64 /usr/local/bin/draftpilot
draftpilot help

Windows x64

For Windows workstations using PowerShell.

Download Windows x64

curl.exe -L -o draftpilot-0.1.3-windows-x64.zip ^
  https://draftpilot-releases.jerriebright.workers.dev/releases/draftpilot-0.1.3-windows-x64.zip
powershell -NoProfile -Command "Expand-Archive -Force draftpilot-0.1.3-windows-x64.zip ."
.\draftpilot-windows-x64.exe help

One-line CLI install: macOS and Linux

curl -fsSL https://jerrinbright.com/draftpilot/install.sh | bash

One-line CLI install: Windows

powershell -ExecutionPolicy Bypass -Command "irm https://jerrinbright.com/draftpilot/install.ps1 | iex"

Launch locally

draftpilot launch \
  --paper /path/to/latex-paper \
  --main-tex main.tex \
  --codebase /path/to/research-code

Launch on a VM

draftpilot launch \
  --host user@vm.example.com \
  --paper /home/user/paper \
  --main-tex main.tex \
  --codebase /home/user/research-code

Update

draftpilot update

Remote VM mode requires DraftPilot on both machines: the local binary opens the tunnel, and the VM binary runs beside the remote paper and codebase.

Release manifest · SHA-256 checksums

Feature Grid

Everything around the paper, in one workspace.

Local-First LaTeX Workspace

Edit papers directly from your local folders with a clean Overleaf-like workflow without losing access to your real development environment.

AI Agents With Real Context

Give AI agents direct access to your manuscript, linked repositories, figures, compile logs, PDFs, and experiment outputs.

SyncTeX Navigation

Jump instantly between LaTeX source and compiled PDF pages for fast editing and debugging.

Local PDF Compilation

Compile papers locally using your existing TeX setup and inspect logs, warnings, and references in real time.

Unified Research Workspace

Manage LaTeX files, figures, bibliographies, PDFs, notes, scripts, and linked codebases from one interface.

Review and Rebuttal Tracking

Highlight TeX or PDF passages, attach comments, track paper TODOs, and jump back to every note while preparing revisions or rebuttals.

Built for Modern AI Research

Designed for ML researchers, robotics labs, systems researchers, PhD students, and AI startups working with complex local workflows.

AI Workflow

Your paper and codebase finally understand each other.

With DraftPilot, AI agents can reason over the entire research workspace instead of seeing a paper-only sandbox.

Trace claims back to implementations.
Locate where figures are generated.
Inspect compile failures and PDF outputs.
Compare equations against code.
Update sections after experiment changes.
Reason across manuscript and implementation together.

This is more than a LaTeX editor. DraftPilot is an AI-native research environment.

Comparison

Why researchers use DraftPilot instead of cloud-only editors

DraftPilot Traditional Paper Editors

Local filesystem accessSandboxed projects

Connected to codebasesPaper-only workflows

AI agent integrationLimited AI context

SyncTeX + local compilationRemote-only compilation

Experiment-aware workflowsDisconnected artifacts

Git-native workflowsPlatform lock-in

Get Started

Write papers where your research already lives.

DraftPilot keeps your manuscript connected to the code, figures, and experiments behind it while enabling powerful AI-assisted workflows locally.

Start with a local paper folder, or connect to the Linux VM where your experiments already run.

Download DraftPilot Quickstart Guide VM Workflow Join Discord

Docs

One clear path for every setup.

Use the quickstart first, then open the guide that matches where your paper and code live.

Quickstart Pick desktop, CLI, or VM mode in under a minute. Desktop App Run DraftPilot locally on macOS or Linux. CLI Launch, update, and script DraftPilot from the terminal. VM Workflow Connect to papers and codebases living over SSH. CoGS Example Concrete command and project setup for the current paper. Releases Binary downloads, checksums, and update metadata. Troubleshooting Fix Gatekeeper, FUSE, SyncTeX, ports, and LaTeX issues.