Bitpull

io.github.SamuealPKPublished by GitHub account SamuealPKv0.1.24Updated Nov 7, 2026

Search shared, signed code repos; use community compute and open AI models; set up BitPull.

Overview

AI-generated overview

Bitpull lets an AI assistant search shared signed code repositories, query open community models, and run jobs on community or own computers.

What you can do
  • Search, read manifests of, and fetch files from shared signed code repositories
  • Ask and run open models, and list the models available on the network
  • Ask and list models run by the user's own group of computers
  • Run sandboxed jobs, check job status, and read the account credit balance
  • Control the desktop app's panel: groups, start/stop/share, login and logout, setup status, lending rules, AI limits, installs and model downloads
When to use it
Useful for developers who want their coding assistant to search published code repositories, try open models, or offload sandboxed jobs to community or personal machines, and who are willing to install the BitPull desktop app or CLI.
Requirements
Runs locally over stdio, typically via npx with Node.js 20 or newer; the full desktop app is installed with the platform install scripts or installers. A BitPull account is needed: the assistant shows a link and code and the user approves login on the website. No API keys or environment variables are declared.
Before you install
Login is approved by the user on the website, so the assistant never handles the password. Lending idle CPU, memory and models is off until enabled and earns credits; plans and pay-as-you-go are paid. Jobs run on other people's computers, which must decrypt them to do the work, so private data should use only the user's own computers. Requests to proprietary models pass through BitPull's server and are not end-to-end encrypted. The app can edit authorized project folders, run commands and push…

Installation

In SourceWeft

  1. Open Bitpull in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

BitPull (bitpull.cloud)

Share code knowledge and computing power with AI tools. BitPull is a desktop app and command-line tool for developers. With it you can:

  • Publish signed, AI-readable code repositories. Claude Code, Codex, Cursor and other AI tools search them through MCP.
  • Lend idle CPU, memory and open AI models (text, vision, images, speech, music) and earn credits. Lending is off until you turn it on, with daily, weekly and monthly limits.
  • Use the community network: run jobs in sandboxes on other people's computers, encrypted on the way, and ask open models.
  • My network: computers you add run your own jobs first, for free. They can also run one larger model together, from the dashboard, and offer it to others for credits.
  • Chat and local APIs: a chat page for every model, and OpenAI, Claude Messages and Ollama compatible endpoints on this computer for your other tools.
  • Projects: a code editor on a folder you authorise, with agents that edit it: file tree, tabs, search, Git panel (push only with your click), Problems, outline, local history, inline edits, @file references, diffs you keep or undo piece by piece, Stop and Undo. Ask always, accept edits or auto: in auto commands run on your computer in the project folder within automatic limits; with community models, and for publishing or risky commands, every command waits for your click.
  • Terminal: a real terminal in the projects view of the desktop app, off until you turn it on. The npm package leaves out its native component to stay small.
  • Costs less: models run on computers people already own. No monthly fee with pay as you go, or plans from €4.50 a month, VAT included.

Not affiliated with other products named BitPull. Website: https://bitpull.cloud · Guide for AI assistants: https://bitpull.cloud/llms.txt · Install data with SHA-256: https://bitpull.cloud/install.json

Install

SystemCommand
macOS / Linuxcurl -fsSL https://bitpull.cloud/install.sh | sh
Windows (PowerShell)powershell -ExecutionPolicy Bypass -c "irm https://bitpull.cloud/install.ps1 | iex"
Only the command line and MCP server (Node.js 20+)npx -y bitpull doctor

The install scripts verify the SHA-256 of the download before installing it, install for the current user and add the bitpull command. Installers are also available at https://bitpull.cloud/download

Use it from your AI tool (MCP)

json
{ "mcpServers": { "bitpull": { "command": "npx", "args": ["-y", "bitpull", "mcp"] } } }

You can also connect it from the app, or with bitpull connect claude-code|codex|cursor|windsurf|gemini|vscode|claude-desktop.

Then ask your assistant to log in. The bitpull_login tool shows a link and a code, and you approve it on the website: the assistant never sees your password.

The tools include:

  • search_repos, read_manifest, fetch_files
  • ask_open_model, use_open_model, list_open_models
  • ask_group_model, list_group_models: the model run by the person's own group of computers
  • bitpull_groups, bitpull_start_group, bitpull_stop_group, bitpull_share_group and the others named bitpull_*: every item of the app's control panel
  • run_job, job_status, my_balance
  • bitpull_login, bitpull_logout, bitpull_setup_status, bitpull_set_lending_rules, bitpull_set_ai_limits, bitpull_install, bitpull_download_model

Command line

bitpull login --device     log in by approving a code on the website (--via github|google)bitpull logout             log this computer out, e.g. to log in with another accountbitpull doctor             what this computer has, required vs optional, and how to add the rest (--json)bitpull status             login, lending state, what this computer offers nowbitpull install lemonade   images, speech and music engine (also: docker, ollama --run)bitpull pull qwen3.5:4b    download an open model to lendbitpull set enabled=on idleMinutes=0 maxCpus=2 keepAwake=on waitMinutes=15bitpull limits day.requests=50bitpull ask "…"            ask an open model on the networkbitpull run python:3.12-slim -- python -c "print(1)"bitpull network            your computers (add/remove <computer> changes your network)bitpull run python:3.12-slim --mine -- python -c "print(1)"

Privacy and safety

  • BitPull reads only the folders you authorise and publishes only what you publish.
  • Jobs travel encrypted and BitPull's server cannot read them. The computer that runs a job reads its content to do the work: for private data, use only your own computers. On your computer they run in sandboxes without internet access or access to your files. More at https://bitpull.cloud/security
  • The computer that runs a job must decrypt it to do the work. For private data use --mine: the job never leaves the computers of your network.
  • The local dashboard and API listen only on 127.0.0.1 and require a secret token.
  • Requests to proprietary models offered by the platform (Claude, GPT, Gemini…) pass through BitPull's server to the provider: they are not end-to-end encrypted.

Terms: https://bitpull.cloud/legal/terms · Privacy: https://bitpull.cloud/legal/privacy

Downloads in this repository

This repository hosts the signed release builds of the BitPull desktop app; the source code is private. Get the latest version from the Releases page, from https://bitpull.cloud/download or with the install scripts above.

  • macOS: open the .dmg and drag BitPull to Applications. The first time, right-click the app and choose Open (the install script avoids this step).
  • Windows: run the -setup.exe installer. If SmartScreen appears, choose More info → Run anyway.
  • Linux: install the .deb or .rpm, or make the .AppImage executable and run it.

Every file's SHA-256 is listed on the release page and in https://bitpull.cloud/install.json.

Source: README.md at commit 29a55da

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

6
  1. v0.1.24LatestNov 7, 2026
  2. v0.1.20Nov 5, 2026
  3. v0.1.19Nov 5, 2026
  4. v0.1.17Nov 4, 2026
  5. v0.1.16Nov 4, 2026
  6. v0.1.15Nov 3, 2026

Score

40% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

    Declare a license in the repository or the registry entry.