MCP100 Assessment

Editorial verdict

Strong knowledge-graph MCP with useful persistent-memory workflows, but current default HTTP exposure and write/delete tools are too risky for Recommended.

Reviewed September 11, 2026

Task Fit

Mapped

Local files and knowledge

Trust

Status warning

mcp_relevance; task_fit; documentation; maintenance_source_health; client_coverage; tool_contract; install_supply_chain; advisory_freshness; adoption_evidence_presence; resource_type_requirements

Maintenance

Date unavailable

No dated repository or package update is currently stored.

Popularity

Signals unavailable

No independently comparable adoption metric is currently stored.

Best for

Self-hosted persistent memory and knowledge-graph workflows
Teams willing to isolate the service behind network/auth controls

Not ideal for

Internet-exposed deployments without a reverse proxy or other access control
Production memory stores that require built-in strong authentication

Avoid if

You would expose the default 0.0.0.0 HTTP endpoint on an untrusted network
You cannot back up or recover graph data before enabling delete/clear tools

Trade-offs

Rich memory tools are valuable but include destructive operations
Multiple LLM/database credentials increase the deployment's secret surface

Evidence scope

Evidence

Current source snapshot

Popularity

No MCP-specific repository metric is stored

No MCP-specific package metric is stored

Repository identity: getzep/graphiti

Stars observation date unavailable · Downloads observation date unavailable

Exact implementation repository metric.

No exact package is documented.

Maintenance

Publisher activity observed September 11, 2026

Current source snapshot

Trust review

Source relationship: First-party MCP

Reviewed September 11, 2026

Sources

Source directory: mcp_server ↗

Evidence-backed setup

Install & Agents

Choose your MCP client. A client connects to the MCP server; the language model is selected inside that client. Only stored, source-backed configurations are shown.

Claude Code

Transport not recorded

Publisher documented

Use official Graphiti MCP server; documented local HTTP endpoint is `http://localhost:8000/mcp/`, with Docker Compose recommended; configure graph DB and LLM credentials.

The publisher provides client-specific setup instructions.

Verify safely: Use Graphiti `http://localhost:8000/health`, then `claude mcp list` if registered.

Uninstall and revoke

Claude Code: `claude mcp remove <name>`; also revoke OAuth/device/API credentials in the vendor account when the vendor provides such a control.

Evidence & methodology

Publisher-documented setup is not the same as MCP100 execution testing. “MCP100 logic checked” means the stored configuration shape was reviewed, not that third-party MCP software was run.

Evidence checked 2026-09-20T00:00:00.000Z

Evidence 1 ↗

Reviewed setup

Type-appropriate, source-backed setup for this mcp server profile.

Transport: HTTP default or stdio
Endpoint: http://localhost:8000/mcp/
Install: Clone getzep/graphiti, configure mcp_server/.env, then docker compose up or uv run main.py.
Verify: Confirm the endpoint is not publicly exposed and inspect tools before using real graph data.
Setup notes: Add network/reverse-proxy auth until a reviewed built-in auth path is confirmed.

Show source-catalog configuration

Use official Graphiti MCP server; documented local HTTP endpoint is `http://localhost:8000/mcp/`, with Docker Compose recommended; configure graph DB and LLM credentials.

Capability

What it does

Strong knowledge-graph MCP with useful persistent-memory workflows, but current default HTTP exposure and write/delete tools are too risky for Selected.

Files & Knowledge

Representative uses

Self-hosted persistent memory and knowledge-graph workflows
Teams willing to isolate the service behind network/auth controls

Important limits

Internet-exposed deployments without a reverse proxy or other access control
Production memory stores that require built-in strong authentication

Tools13 toolsView tools

add_memory
add_triplet
search_nodes
search_memory_facts
summarize_saga
build_communities
get_episode_entities
delete_entity_edge
delete_episode
get_entity_edge
get_episodes
clear_graph
get_status

Source: Official documentation

Context Footprint

Not measured yet

Estimated token footprint of the MCP tool schemas exposed to the agent.

Reviewed access

Access & risk

Review the exact access boundary before connecting this MCP to accounts, credentials, or sensitive data.

Authentication: Current MCP server docs do not establish built-in HTTP authentication as a default control; deployments should rely on local access or external network/reverse-proxy controls.
Permissions: Tools include add/search plus delete_episode, delete_entity_edge, and clear_graph. Database privileges are those of the configured Neo4j/FalkorDB credentials.
Data handling: Stores knowledge-graph episodes/entities/facts and uses configured LLM/embedder providers during ingestion/search. Anonymous telemetry can be disabled.
Cost: Open-source server; database and LLM/embedder provider costs apply.
Risk and cautions: High. Default network exposure plus destructive graph tools and backend/LLM credentials create material unauthorized-access and data-loss risk.

Related reviewed MCPs

Other options

Codebase Memory MCP

MCP100 Recommended

Codebase knowledge graph for AI agents — 159 languages, sub-ms queries, 99% fewer tokens.

Choose this instead if

Choose Codebase Memory MCP instead if repository indexing, symbol search, call-path tracing, impact analysis, and architecture queries across many languages.

Recommended after reviewed 2026-09-25 admission dossier: all_current_selected_gates_pass.

Compare with Graphiti MCP View Codebase Memory MCP details

io.github.blazickjp/arxiv-mcp-server

MCP100 Recommended

Recommended local research MCP for arXiv literature search, bounded full-text/LaTeX reading, citation traversal, local paper storage, and research alerts.

Choose this instead if

Choose io.github.blazickjp/arxiv-mcp-server instead if researchers who want focused arXiv literature workflows with local storage and bounded reads.

Recommended local research MCP for arXiv literature search, bounded full-text/LaTeX reading, citation traversal, local paper storage, and research alerts.

Compare with Graphiti MCP View io.github.blazickjp/arxiv-mcp-server details

MCPVault for Obsidian

MCP100 Recommended

Give MCP clients scoped access to a local Obsidian vault.

Choose this instead if

Choose MCPVault for Obsidian instead if obsidian users who want local search and supervised note maintenance.

Give MCP clients scoped access to a local Obsidian vault.

Compare with Graphiti MCP View MCPVault for Obsidian details

Read publisher documentation

Publisher documentation

MCP100 source-backed summary

A concise MCP100 summary built from the reviewed publisher sources. It is not a publisher README; use the original link for the latest full documentation.

View original documentation ↗

Graphiti MCP

Strong knowledge-graph MCP with useful persistent-memory workflows, but current default HTTP exposure and write/delete tools are too risky for Selected.

What it can help with

Self-hosted persistent memory and knowledge-graph workflows
Teams willing to isolate the service behind network/auth controls

Important limits

Internet-exposed deployments without a reverse proxy or other access control
Production memory stores that require built-in strong authentication

Loading optional publisher documentation…

FAQ

What does Graphiti MCP do?

Strong knowledge-graph MCP with useful persistent-memory workflows, but current default HTTP exposure and write/delete tools are too risky for Selected.

How does Graphiti MCP authenticate or get access?

Current MCP server docs do not establish built-in HTTP authentication as a default control; deployments should rely on local access or external network/reverse-proxy controls.

What permissions does Graphiti MCP have?

Tools include add/search plus delete_episode, delete_entity_edge, and clear_graph. Database privileges are those of the configured Neo4j/FalkorDB credentials.

What are the main risks of using Graphiti MCP?

High. Default network exposure plus destructive graph tools and backend/LLM credentials create material unauthorized-access and data-loss risk.

What does Graphiti MCP cost to use?

Open-source server; database and LLM/embedder provider costs apply.

Is this an official MCP?

reviewed-detail:graphiti-mcp is listed as First-party MCP in MCP100’s current evidence.

Current official resources: