forest

FOREST is a research agent for long-horizon science, where evidence gets lost, errors compound, and goals drift. It explores ideas, reads literature, forms hypo

What is forest?

FOREST is a research agent for long-horizon science, where evidence gets lost, errors compound, and goals drift. It explores ideas, reads literature, forms hypotheses, runs real experiments, analyzes results, revises paths, and builds a full paper—all in an editable graph you can inspect, branch, redirect, with claims and figures linked to evidence

Overview of forest

What it is

FOREST is an end-to-end research agent framework for carrying a scientific project from an early question to a finished manuscript. It is designed for long-horizon science, where evidence gets lost, errors compound, and goals drift. FOREST represents the research process as an editable research graph, connecting ideas, hypotheses, experiments, results, decisions, and paper sections.

Key features

  • Explores and compares research directions, searches the literature, and turns promising ideas into testable hypotheses.
  • Designs experiments with baselines, controls, metrics, ablations, and explicit success/failure conditions.
  • Runs real experiments by writing and executing research code, launching training and evaluation jobs, and collecting logs and metrics.
  • Analyzes results, compares competing explanations, identifies failures, and decides whether to extend, revise, or drop a direction.
  • Keeps the research path editable, allowing intervention at any point: edit nodes, change experiments, revisit assumptions, or branch the graph.
  • Continues across execution rounds with durable checkpoints, sharing resource accounting across rounds and branches.
  • Builds publication-ready figures and tables tied to underlying runs, and writes a full manuscript with traceable claims.

Use cases

FOREST is suited for scientific research projects that require long-horizon exploration, iterative experimentation, and evidence-driven decision-making. It can be used in domains such as PDE optimization, time-series forecasting, and medical image classification, as demonstrated in the README's evaluation table. Researchers can start with a broad question or unfinished idea and let FOREST explore directions, run experiments, and produce a paper.

Getting started

Requires Python 3.11+ and Node.js 22+ for the Web workspace. Clone and install: bash git clone https://github.com/mikamikasuki/FOREST.git cd FOREST For Web workspace (macOS/Linux): bash python3.11 scripts/install.py .venv/bin/forest serve Open localhost:8000, configure a model provider in Settings, and create a project with a research goal and spending budget. For CLI/headless: bash python3.11 scripts/install.py --headless .venv/bin/forest serve --headless --data-dir ./var Then use client commands like .venv/bin/forest --help, init, and status.

License and pricing

The license is Apache-2.0, and the pricing type is open source. No pricing information is stated beyond that.

Based on the project's public README and repository metadata.

Add this badge to your README

Featured on YunoDB Badge styles & help

Maintain this project? Show that it is featured on YunoDB.

Markdown[![Featured on YunoDB](https://yunodb.com/badge/mikamikasukiforest.svg)](https://yunodb.com/tools/mikamikasukiforest?utm_source=github&utm_medium=badge)
HTML<a href="https://yunodb.com/tools/mikamikasukiforest?utm_source=github&utm_medium=badge"><img src="https://yunodb.com/badge/mikamikasukiforest.svg" alt="Featured on YunoDB" height="20"></a>