Getting Started

What Memgraph Ingester Does

Memgraph Ingester indexes source files into Memgraph so agents can work with a real code graph instead of a loose pile of text chunks.

It captures structure such as files, packages, types, methods, fields, inheritance, implementations, annotations, decorators, and best-effort call relationships. It also creates optional RAG chunks that point back to source graph records, so semantic discovery can be verified through exact graph lookups.

Why This Saves Agent Context

Memgraph Ingester helps agents start from structure instead of raw file scanning. They can ask the graph for the likely files, symbols, callers, callees, and semantic anchors before spending context on source reads.

In a six-task comparison, Memgraph-backed runs used 41% fewer tokens overall, with the strongest real task saving 62%. They also needed 45% fewer tool calls and 44% fewer opened files.

Metric	With Memgraph	Without Memgraph	Reduction
Tokens used	354,970	605,578	41% fewer
Tool calls	92	168	45% fewer
Files opened	59	105	44% fewer

These measurements were performed on this codebase. At the time of measurement, cloc reported 28,588 code lines across tracked Java, JavaScript, Python, and TypeScript sources: 26,006 Java, 1,863 JavaScript, 645 Python, and 74 TypeScript.

Measured task scope:

Task	Description
A	Provider onboarding
B	Write-path performance
C	Refactor impact
D	Stale-snippet triage
E	Stale-snippet fix
F	CI triage

Quality remained high across the measured tasks, while the graph-guided runs spent less effort on discovery and kept more context available for understanding, editing, and verification. In the measured set, Memgraph also avoided the false positive seen without the graph.

Supported inputs:

Language path	Coverage
Java	First-class parser with JavaParser and optional classpath-based symbol resolution
JavaScript and TypeScript	First-class parser with managed or system Node.js runtime
Python	First-class parser with managed or system CPython runtime
Other ctags-detected languages	Structural file, owner, type, method, and field inventory

Requirements

For normal use you need:

• A Memgraph instance with Bolt enabled.
• A Memgraph Ingester executable or shaded JAR.
• memgraph-ingester-mcp when you want agents to use high-level graph tools.

Native binaries do not require Java, Node.js, Python, or ctags for normal managed-mode parsing. The shaded JAR requires a Java 25 JRE.

1. Start Memgraph

docker run -p 7687:7687 -p 7444:7444 --name memgraph memgraph/memgraph-mage:3.9.0

Memgraph Bolt will listen on bolt://localhost:7687.

2. Install Memgraph Ingester

With Homebrew:

brew tap ousatov-ua/memgraph-ingester
brew install memgraph-ingester

Or download a release asset from GitHub Releases.

For the shaded JAR:

java -jar /path/to/memgraph-ingester.jar --help

In the examples below, replace memgraph-ingester with either the native executable or java -jar /path/to/memgraph-ingester.jar.

3. Ingest a Project

Run this from the repository you want to index:

memgraph-ingester \
  --source . \
  --bolt bolt://localhost:7687 \
  --project my-project \
  --wipe-project-code \
  --init-instructions \
  --apply-schema

This command:

• Applies graph schema setup.
• Clears only code graph data for my-project.
• Scans files under the current directory.
• Writes graph nodes and relationships to Memgraph.
• Installs agent instructions for the project.

Add memories when you want durable project knowledge too:

memgraph-ingester \
  --source . \
  --bolt bolt://localhost:7687 \
  --project my-project \
  --with-memories \
  --wipe-project-code \
  --init-instructions \
  --apply-schema

4. Verify the Graph

With memgraph-ingester-mcp, ask your agent to call server_status for the project.

DIY. With mgconsole:

mgconsole --host localhost --port 7687 --output-format=csv

MATCH (p:Project {name: 'my-project'})-[:CONTAINS]->(l:Language)-[:CONTAINS]->(c:Code)
RETURN p.name, l.name, c.sourceRoots, c.lastIngested
ORDER BY c.lastIngested DESC;

5. Re-ingest While You Work

Refresh the graph after code changes:

memgraph-ingester \
  --source . \
  --bolt bolt://localhost:7687 \
  --project my-project

To keep the graph fresh continuously:

memgraph-ingester \
  --source . \
  --bolt bolt://localhost:7687 \
  --project my-project \
  --watch

Safety Notes

Project data is scoped by --project. --wipe-project-code deletes only the named project's code graph. Memory is protected unless you explicitly pass --wipe-project-memories.

--wipe-all deletes all Memgraph data. Use it only when you intentionally want an empty database.