# Memgraph plugins

The [Memgraph marketplace](https://github.com/memgraph/marketplace) has the
plugins that help AI coding agents such as Claude Code and Codex build with
Memgraph:

- **`memgraph`** gives the agent what it needs to build with Memgraph, in one
  install:
  - **[Agent Skills](#available-skills):** Memgraph-specific knowledge the
    agent loads when a task needs it, for example when writing Cypher,
    designing a graph model or running a MAGE algorithm.
  - **The [Docs MCP server](https://memgraph.com/docs/getting-started/coding-with-ai/docs-mcp):** the
    agent looks up the current docs instead of answering from memory.
  - **The [Memgraph MCP server](https://memgraph.com/docs/ai-ecosystem/integrations/mcp):** the agent
    reads your schema and runs queries on your Memgraph.
  - **An onboarding:** ask "Get started with Memgraph", and the agent starts
    Memgraph and Memgraph Lab in Docker, loads a sample dataset you pick, and
    runs a first query and a graph algorithm.
- **`context-graph`** gives your agents [AI memory](#ai-memory) across
  sessions. It's optional, so nothing is recorded unless you install it.

With other agents, such as Cursor, [install the skills
alone](#other-agents-skills-only).

## Install in Claude Code and Codex

Add the Memgraph marketplace, then install the `memgraph` plugin.

In Claude Code:

```plaintext
/plugin marketplace add memgraph/marketplace
/plugin install memgraph@memgraph
```

In Codex:

```bash
codex plugin marketplace add memgraph/marketplace
codex plugin add memgraph@memgraph
```

Then, with Docker running, ask your agent "Get started with Memgraph". In
Codex, start the agent with `codex -a on-request`, so it can ask to run Docker
outside its sandbox.

The Memgraph MCP server runs with `uvx`, so it needs
[uv](https://docs.astral.sh/uv/). It connects to `bolt://localhost:7687`; to
use another Memgraph, set `MEMGRAPH_URL` before you start the agent.

If you added `memgraph/skills` as a marketplace before, remove it first
(`/plugin marketplace remove memgraph` in Claude Code, `codex plugin marketplace
remove memgraph` in Codex), then add `memgraph/marketplace`.

## Other agents: skills only

The skills follow the open [Agent Skills
specification](https://agentskills.io/specification), so they work with any
agent that supports it. To give the agent the docs as well, add the [Docs MCP
server](https://memgraph.com/docs/getting-started/coding-with-ai/docs-mcp).

### With the skills CLI

Install all skills with the [`skills` CLI](https://skills.sh):

```bash
npx skills add memgraph/skills
```

To install a single skill, pass its name:

```bash
npx skills add memgraph/skills --skill memgraph-cypher-syntax
```

The command asks which skills and agents to install. To install without
prompts, for example from a script or from inside a coding agent, name the
skills and the agent and pass `--yes`:

```bash
npx skills add memgraph/skills --skill '*' --agent claude-code --yes
```

`'*'` installs all skills. Replace `claude-code` with your agent's name in the
`skills` CLI, such as `cursor` or `codex`.

### Manual install

Clone the repository and symlink the `skills` directory into the location your
agent scans, for example `.claude/skills` for Claude Code or `.cursor/skills`
for Cursor:

```bash
git clone https://github.com/memgraph/skills.git ~/.memgraph-skills
cd /path/to/your/project
ln -s ~/.memgraph-skills/skills .claude/skills
```

## AI memory

The `context-graph` plugin lets your agents remember what they did in earlier
sessions, with [AI memory](https://memgraph.com/docs/ai-ecosystem/ai-memory). It needs the Context Graph
CLI, so install it with its setup: one command installs the CLI, starts its own
Memgraph and connects your agents, plugin included. See [Get started with
Context Graph](https://memgraph.com/docs/ai-ecosystem/ai-memory/get-started).

## Available skills

| Skill | Use it when you need to |
|-------|-------------------------|
| `memgraph-cypher-syntax` | Write, fix or optimize Cypher for Memgraph, including BFS/DFS/WSP lambdas, text search and vector search |
| `memgraph-model-graph-data` | Design a graph data model, decide between properties and relationships, or import CSV/Parquet data |
| `memgraph-import-data` | Load or migrate data from CSV, Parquet, JSON, CYPHERL, relational databases or Neo4j, and verify the import |
| `memgraph-indexes-and-constraints` | Create and manage indexes, constraints and enums, and run `ANALYZE GRAPH` |
| `memgraph-database-configuration` | Configure triggers, transactions, isolation levels, storage modes, snapshots, WAL and memory |
| `memgraph-run-mage-algorithms` | Run MAGE algorithms such as PageRank, community detection, centrality, embeddings and path traversals |
| `memgraph-graph-rag` | Build a GraphRAG system: schema design, ingestion, hybrid retrieval and agent tool contracts |
| `memgraph-python-query-modules` | Build custom query modules in Python with the `mgp` API |
| `memgraph-cpp-query-modules` | Build custom query modules in C++ with the `mgp.hpp` API |
| `memgraph-rust-query-modules` | Build custom query modules in Rust with the `rsmgp-sys` API |
| `memgraph-mgconsole-cli` | Connect with `mgconsole`, run queries and configure output |
| `memgraph-lab-write-gss` | Style graph visualizations in Memgraph Lab with Graph Style Script (GSS) |
| `memgraph-brand-ui` | Generate UIs, dashboards and diagrams in the Memgraph brand style |
| `memgraph-start` | Get started: run Memgraph and Memgraph Lab in Docker, load a sample dataset, and run a first query and a graph algorithm |

The full content of each skill is in the
[memgraph/skills](https://github.com/memgraph/skills) repository.