OpenMake
ENKOJAZHDE
Live Demo

OPEN-SOURCE PROJECT

Run your own AI workspace.Build it together.

OpenMake connects models, tools, and agents in infrastructure you control. The MIT-licensed project is open to use, inspect, and improve with us on GitHub.

Build with us on GitHubTry the live demo
Read deployment docs

Build with us on GitHub, try the live demo, or run it yourself — three separate doors.

REAL APP CAPTURE / DEEP RESEARCH

From a question to a report with sources.

OpenMake deep research screen showing collected sources and a completed EV battery report in KoreanView full-size capture
Captured from the Korean app: an EV battery research request and its report. Your results will differ.See how research works

MODEL CONNECTIONS

Choose where your models run.

How to connect models

RUN IT YOURSELF

One command to start on macOS or Linux.

The script selects the installer for your OS and guides the remaining setup. On Windows, run it inside WSL2.

Read the self-hosting guide
shell
curl -fsSL https://raw.githubusercontent.com/openmake/openmake_llm/main/install.sh | bash

AN OPEN PROJECT

The problem we are working on together.

As models and tools multiply, it gets harder to see where work runs and where data goes. We are building a workspace that puts those choices in the user's hands. The code, development notes, and unfinished roadmap are public so experience and disagreement can improve the next decision.

The openmake_llm project is led by maker riskpw, with development support from developer rocky. OpenMake Team is the community around the work. You can join through a bug report, a clearer explanation, or a focused code change.

02

Improve the explanations

Read the README and setup guide. Correct an error, add an example, or improve a translation, then send a pull request.

Read the contribution steps
PUBLIC PROJECT RECORD

See what changed and what comes next.

01Latest development log02Current and planned work

ONE OBSERVABLE WORKFLOW

Follow the work from request to report.

The app capture above shows a completed EV battery research task. This is the path you can inspect in the current software, with results that depend on the connected model and sources.

  1. 01

    Ask

    Start a research task from the chat workspace.

  2. 02

    Trace

    Review source collection and the progress of the research.

  3. 03

    Review

    Read the report and check its cited sources before using it.

Explore the workflow in the manual

ENGINEERING STORIES

Follow how we build it

Weekly changes, real problems, design decisions, and evidence from building OpenMake.

View all engineering notes →
01 / WEEKLY LOG

2026‑W43: Colima Runtime Unification, Live Terminal Streaming, and Robust Task Cancellation and Recovery

At October 24, 19:30 KST, main 6dfd605a contains 367 commits and 16 tagged releases (1.106.2 through 1.115.1). Unifies macOS runtime exclusively under Colima with isolated Playwright sandboxes, adds a web live terminal streaming panel with auto-follow scroll, connects an eight-stage verification and delivery pipeline, solidifies security floors including NULL-as-inactive API keys (S5), propagates taint boundaries across chat subagents, routes immediate cancellation signals through research and MCP meta boundaries, preserves chat session continuity after first-turn aborts, and introduces watchdog stale pending lease detection with resolved notifications. W43 snapshot distinguishing source records from fresh operational evaluation.

  • Weekly log
  • Vibe coding
  • Self-hosted AI
Read the engineering note

JOIN THE PROJECT

Find a small place to start.

Look at an open issue or try the software. A useful report or clearer document helps as much as a code change.

Start with an open issueTry the live demo