Skip to content
GraphForge v0.6.0

GraphForge

Build a graph, ask a question, and keep what you learn.

GraphForge records items and their connections: for example, which paper cites another, or which interview excerpt informed a finding. You can ask questions about those connections and keep the evidence behind an answer. It runs on your computer. The core audience is a nontechnical analyst working with an agent; v0.6.0 initially serves technical analysts and integrators more completely. Who GraphForge is for names two jobs it completes today and when an embedded engine does not fit.

These docs target v0.6.0, which is pre-v1. Installation explains release availability, setup, platforms and limits. Only supported current-format projects can be opened; backward compatibility and migration are not guaranteed.

Start with the public Hub

Section titled “Start with the public Hub”

The coordinated launch path is browse the production Hub → install the exact candidate CLI → clone → reopen/query → continue locally → publish back → discover/clone the new Version. The Hub guide explains the prerequisites and current qualification boundary. The intended production fixture is not yet available through discovery; this journey is not yet qualified. Local graph use below remains available.

Basic: understand and use a graph

Section titled “Basic: understand and use a graph”

Start here if you are doing your first research project. No programming knowledge is needed to understand the lessons. Running the examples uses a coding agent or a technical helper for setup.

  1. Your first graph: connect three papers and check a citation count.
  2. Work with an agent: ask it to run the example and inspect the actual answer.
  3. Your first mixed-methods project: connect survey responses and interview excerpts, challenge an explanation, then save and retrieve a bounded conclusion.

Basic graph use is a complete outcome. You do not have to learn collaboration, research history, or every other feature.

Advanced: work directly with the tools

Section titled “Advanced: work directly with the tools”

The Advanced introduction assumes basic Python, database, and terminal skills. It explains GraphForge-specific ideas as you need them; expert knowledge is not a prerequisite.

Your task Start here
Run code and inspect results Use a notebook
Save a graph and return later Query, analyze, and save
Write queries or load records Cypher · Graph construction
Retain a challenged hypothesis and its evidence Record and revisit an inquiry
Keep an exact research state Keep research history
Move or share a saved project Portable projects
Call GraphForge from an application Integrate GraphForge

Choose the task you need. Advanced is a collection of optional paths, not a sequence you must finish before using the product.

When GraphForge fits

Section titled “When GraphForge fits”

Use it to inspect relationships in a citation network, connect observations to sources, analyze a dependency graph, or retain the reasoning behind an inquiry. Start with the data you have; an ontology is not required to create a graph.

For a centrally hosted, high-throughput application serving many concurrent users, evaluate an operational database against that workload. GraphForge’s scale guidance explains the difference between measured workloads and a general size promise.

Durable projects require a supported local filesystem. In-memory work ends when the process or notebook kernel closes. Read the storage requirements before saving work you intend to keep.

Reference and specialist material

Section titled “Reference and specialist material”

Architecture and implementation contracts are specialist references, outside the Basic and Advanced learning paths.

Measured scale, with its workload, host, and limits, is in claims and evidence. For help, see Get help and report a problem: ask in GitHub Discussions, or file a GitHub issue with a minimal reproduction. GraphForge is available under the Apache License 2.0.