Neo4j MCP Integrations
Infrastructure · tested 2026-08-23 · re-test due 2026-11-23 · by the Hlido desk, not the vendor
In short: Vendor-official MCP integration guidance for Neo4j that is unusually candid about the protocol's unsolved problems — strong on trust, thin on the servers themselves.
4 PASS · 1 FAIL of 5 public-surface claims
Quick answer
Neo4j MCP Integrations scores 74/100 (STEADY) on Hlido’s independent, hands-on test (reviewed 2026-08-23). STEADY (74) because vendor-official documentation from an established database company carries a high continuity floor, the MCP explanation is accurate and non-promotional, and the candid statement of the protocol's open Pricing was not findable on the public surface when tested.
This is Neo4j's own developer-centre page for Model Context Protocol integrations against the graph database. As a surface it does two things well and one thing incompletely. It explains MCP accurately — host, servers, clients, tools, resources, prompts — and situates it in the wider ecosystem, naming the cloud and framework adopters (Google, Microsoft, AWS, OpenAI, the Docker MCP Catalog, LangChain, CrewAI, Pydantic.AI, ADK) rather than implying Neo4j invented the category. More notably, it states plainly that MCP still has open issues around security, discovery, trusted deployment and permission control. A vendor page that names the unsolved risks in the protocol it is promoting is a meaningfully better trust signal than one that does not, and it is rare. What the surface does not do is give a buyer or an agent a crisp inventory of what the Neo4j MCP servers actually expose: which tools, what scoping, read-only or read-write, and what happens when an agent gets Cypher access to a production graph. The page is a developer guide sitting inside a large documentation estate, so the reviewed surface is orientation rather than product specification. That is a fair thing for it to be — but it means the evaluation stops one level short of the thing that matters most, which is the security boundary around graph write access.
Why STEADY
STEADY (74) because vendor-official documentation from an established database company carries a high continuity floor, the MCP explanation is accurate and non-promotional, and the candid statement of the protocol's open security problems is a genuine trust signal. Held below the 80s because the reviewed surface is an orientation guide rather than a server specification: the tool inventory, scoping model and write-access boundary — the details that decide whether this is safe to give an agent — are not resolved here.
Public-surface checklist
- PASS Homepage loads (required)
- PASS Primary value prop (required) — Model Context Protocol (MCP) Integrations for the Neo4j Graph Database
- PASS Cta present (required) — Get Started Free
- FAIL Pricing or access — No pricing or access terms on the reviewed guide surface; a 'Get Started Free' path exists but terms are elsewhere
- PASS Evidence or demo — 3 screenshots captured; guide links to comprehensive overview with examples
What we saw
3 screenshots captured by the Hlido engine during the reviewed run (run-c703bb83d24db4c2-neo4j-com). Our own captures — not vendor marketing material.
What it does well
- Names MCP's open problems (security, discovery, trusted deployment, permission control) rather than glossing them
- Accurate, non-promotional explanation of the MCP host/server/client architecture
- Situates Neo4j within the ecosystem instead of overclaiming ownership of the category
- Vendor-official surface from an established database company — high continuity floor
- Links onward into GraphRAG guides, examples and the wider GenAI ecosystem
What it fails at
- No inventory of the actual tools the Neo4j MCP servers expose
- Read-only versus read-write scoping is not resolved on this surface
- No security guidance specific to giving an agent Cypher access to a production graph
- Cookie-consent interstitial is the first content encountered — a real friction for automated readers
- Reviewed surface is a guide, not a product page, so version and release state are absent
Best for
- Teams already on Neo4j evaluating whether to expose the graph to an agent
- GraphRAG builders who need the vendor's framing before choosing a server
- Anyone wanting an honest, vendor-authored primer on MCP's current limitations
Not recommended for
- Buyers needing a concrete tool inventory and permission model before adoption
- Non-Neo4j graph stacks
- Agents attempting to auto-discover a callable endpoint from this page — it is prose, not a manifest
Pricing & access
- Pricing findable on the public surfaceFAIL No pricing or access terms on the reviewed guide surface; a 'Get Started Free' path exists but terms are elsewhere (tested 2026-08-23)
Derived from Hlido-held evidence only (engine checklist + editorial text); quotes are verbatim from the scorecard; not vendor-supplied; re-derived daily. Verify current prices on the vendor's pricing page. Last verified 2026-08-23.
Compared to
-
@langchain/mcp-adapters
expose-vs-consume
LangChain's adapters are framework-side plumbing for consuming any MCP server; this is a database vendor's guidance for exposing one. Read the Neo4j page to decide what to expose, the adapters to wire it in.
-
BioMCP
specification-clarity
BioMCP is a domain-specific server with a concrete published tool surface; the Neo4j page is general integration guidance without one. BioMCP is easier to evaluate; Neo4j has the stronger vendor floor.
Agent relevance
API MCP SDK
Agentic-Commerce Readiness 67/100 · INTEGRABLE
Independent readiness for agent delegation & transaction. How it’s scored · check live
Neo4j publishes MCP server integrations for the graph database, and this guide is the vendor's entry point to them. The page itself is prose rather than a machine-readable manifest, so an agent must follow onward links to reach an installable server; the tool surface and permission scoping are not discoverable from here.
Agent-friendly score: 6/10
Score over time
The longitudinal record — every point is the score as published on that date. Raw series.
Evidence
- Vendor-official Neo4j developer guide for MCP integrations — source (2026-08-23) verified
- States open issues remain in MCP around security, discovery, trusted deployment and permission control — source (2026-08-23) verified
- Names major cloud and framework MCP adopters — source (2026-08-23) verified
- No tool inventory or permission model published on this surface — source (2026-08-23) verified

