Create Context Graph CLI
A Neo4j Labs CLI generator: choose an industry and an agent framework to get an app with a knowledge graph, chat and MCP.
- 747 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- The generator connects a Neo4j database, model keys and can read Claude Code session history.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/create-context-graphDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The generator connects a Neo4j database, model keys and can read Claude Code session history.
This is third-party code. Review the repository files before installing.
What it does
Create Context Graph asks questions in a wizard and builds a ready project with a context graph. The output is a FastAPI backend with an AI agent for the chosen industry, a Next.js frontend with streaming chat and graph visualization, a Neo4j schema and demo data. The agent framework is picked with `--framework`: `strands`, `pydanticai`, `claude-agent-sdk`, `openai-agents`, `langgraph`, `crewai`, `google-adk` or `anthropic-tools`. The `--with-mcp` flag adds an MCP server config for Claude Desktop. The Claude Code connector reads session history from `~/.claude/projects/` and extracts decisions and preferences.
Who it is for. Developers of agent applications who want a working scaffold with Neo4j graph memory instead of an empty template.
Good fit when
- A prototype agent for healthcare, finance or another industry is assembled in an evening.
- Several agent frameworks need comparing on one dataset.
- Claude Code session history is turned into a graph of decisions and preferences.
- The graph is exposed to Claude Desktop through a generated MCP config.
Not a fit when
- A supported product is required rather than an experimental project.
- Python 3.11 and Node.js 18 cannot be installed.
- The graph and memory should run with neither a NAMS key nor your own Neo4j.
Example request
Generate a `my-app` project for the healthcare industry on `pydanticai` with local Neo4j and demo data, then list the run steps.Limitations
Neo4j Labs marks the project as experimental and not officially supported: help comes through GitHub Issues and the forum. The last commit in the repository is from October 2026. Python 3.11, Node.js 18 and an agent key such as `ANTHROPIC_API_KEY` are required. Memory storage is either the hosted NAMS service with a key or your own Neo4j 5 or newer. SaaS data connectors need your own access to GitHub, Slack, Gmail and other services.
How to disable. Stop the generated app and delete its folder. Remove the Neo4j container, NAMS keys and imported data separately, and take the MCP entry out of the Claude Desktop config.
FAQ
How do I start the wizard?
With `npx create-context-graph` or `uvx create-context-graph my-app ...` plus the flags you need.
How do storage modes differ?
By default memory is hosted in NAMS and the graph starts empty. With `--self-hosted` your own Neo4j is used and `make seed` loads demo data.
How do I add MCP for Claude Desktop?
With the `--with-mcp` flag at generation: the project gets an `mcp/` folder with the config.
What does the Claude Code connector do?
It reads local session history from `~/.claude/projects/` with no keys and extracts decisions and preferences. Secrets are redacted before storage.
Which industries are available?
The wizard offers industries such as healthcare, and a custom domain can be described in plain text so the model generates an ontology.
Discussion
No questions or reviews yet.
Related
CLI toolsNine markdown files turn a Claude Project into a multi-agent context engine with plan checks and an execution trace
Agent memory in the Memgraph graph database: sessions, tool calls and notes that an agent can recall in later sessions.
ContextLattice: a local context layer for agents
ContextLattice
A local memory and context layer: a session's goal, decisions and evidence are gathered into a compact packet an agent reopens
BrainAPI turns text into a knowledge graph on your server and serves agents over MCP, REST and a console with fact-path answers
UltraContext collects Claude Code, Codex and OpenClaw sessions in one place so agents can see each other's context.
A template for Claude Code projects: parallel agent swarms, workflow skills, safety hooks and rules in one install.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Create Context Graph (this card) | no | 747 | ||
| Neo4j Labs MCP Servers | no | API key | 989 | a set of MCP servers for Neo4j access; this generates a whole app with an agent, chat and graph |
| Graphiti MCP | no | API key | 32,776 | a ready temporal-graph MCP server; here a wizard builds a project for the chosen industry and framework |
README in short
The README shows a quick start via `uvx` or `npx`, industries and frameworks, SaaS data connectors, the generated project structure and a flag reference.