OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Research Agents

MISAKA

MISAKA is an open-source Python multi-agent research system for humanities and social-science inquiries. A coordinator branches research questions among specialist agents, gathers their reports, and subjects conclusions to review within a directed acyclic graph.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

MISAKA — OpenViST project profile in Research Agents.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use MISAKA when an inquiry benefits from parallel specialist research, explicit review of assumptions and gaps, and a record of how conclusions changed. The user approves plans and can inspect, halt or resume agent runs.

Who it is for

  • Humanities researchers planning literature-informed inquiries.
  • Social-science researchers who want to coordinate parallel specialist investigations.
  • Researchers who need to review how conclusions developed and trace citations to source locations.

Documented uses

  • Break a research question into approved branches and assign specialist work.
  • Collect and synthesize reports from agents with different specialties, skills and models.
  • Challenge conclusions, surface coverage gaps and consider new research directions.
  • Trace citations to page and character offset and preserve plans, critiques and conclusion versions in project files.
Before you choose

Requirements & limits

The README documents macOS, Linux and Windows support on x86_64 or arm64, with macOS recommended. Setup requires uv, Git, ripgrep, fd, poppler and model-provider access through an API key or a ChatGPT or GitHub Copilot subscription. Claude sign-in is also described, with usage billed by Anthropic per token as extra usage. The project is installed from its GitHub repository; the PyPI package named misaka is identified

  • MISAKA is a local project-oriented system rather than a hosted research workspace: the README directs users to create a research folder and run the tool there.
  • The user approves plans, can enter agent panes, and controls when project changes are committed.
  • Research branches inherit earlier reasoning; converging lines are compared, and shared questions can be researched once.
  • The process includes coverage tables, literature scans, critique and citation traceability.
  • The README names Claude, GPT, Gemini and local models for specialists, but setup still requires access to a model provider.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install MISAKA and create a research workspace

Set up MISAKA in a project folder and complete its initial configuration.

Before you start

  • macOS, Linux, or Windows (x86_64 or arm64); macOS is recommended.
  • Install uv, git, ripgrep, fd, and poppler.
  • Access to a model provider through an API key, ChatGPT subscription, or GitHub Copilot subscription.

Steps

  1. Install MISAKA from its GitHub repository with the command below.

    Documented command
    uv tool install "misaka[providers] @ git+https://github.com/Luciole-Studio/Misaka-Agent.git"

  2. Create a working folder under Documents and change into it; do not run MISAKA in the root directory.

  3. Run misaka setup to sign in, choose a model, and create the first two Sisters.

Expected outcome

MISAKA is installed and initially configured in a research project folder.

Guide limitations

  • The PyPI package named misaka is an unrelated project; install from the repository as shown.
  • Claude account usage is billed by Anthropic per token as extra usage.
  • ocrmypdf, DjVuLibre, and LibreOffice are optional and are not included in the prebuilt archives.
Documentation 1 ↗Documentation 2 ↗
Source-based practical guide

Start a research session from a configured workspace

Start MISAKA and initiate its documented `/research` workflow.

Before you start

  • MISAKA is installed and configured, with a model selected and the first two Sisters created.
  • A working folder under Documents; do not run MISAKA in the root directory.

Steps

  1. Change to the research project folder.

  2. Run misaka to start MISAKA.

  3. Type /research to begin the research workflow.

Expected outcome

The research workflow starts, with MISAKA coordinating a question, specialist research branches, reports, and review.

Guide limitations

  • The research plan proceeds only after user approval.
  • A research run takes one question and explores different ways it could be answered.
Documentation 1 ↗Documentation 2 ↗
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
7 October 2026
Original website title
Luciole-Studio/Misaka-Agent · Official homepage title · retained in its source language
How we review sourcesSuggest a correction