Official reference MCP server that gives Claude persistent memory through a local knowledge graph of entities, relations, and observations, stored in a file on your own machine.
by Model Context Protocol · submitted by glorydavid03023·added 2026-06-03·
Persists remembered information to a local file on disk (its location can be set with MEMORY_FILE_PATH); manage and review that file as it accumulates data., Runs locally as an stdio process and makes no external network calls; the knowledge graph stays on your machine.
Privacy notes
Details Claude chooses to remember are written to and read from the local memory file, so it can retain personal or project information across sessions., Review or clear the memory file if it should not retain sensitive data, and use MEMORY_FILE_PATH to control where it is stored.
Author
Model Context Protocol
Submitted by
glorydavid03023
Claim status
unclaimed
Last verified
2026-06-03
Decision playbook
Review trust signals before you adopt
Signals are present but mixed. Use the checklist below to confirm the source and operational safety for your environment.
Compare context
Selected
0
Current score
78
Baseline
—
Delta
No baseline selected
No major trust-signal divergence detected in the current selection.
Source and provenance checks
Complete
Confirm ownership and provenance before trusting install instructions.
Source link availableRequired
Open the canonical repository and verify ownership.
Done
Source provenance statusRequired
Marked as source-backed.
Done
Metadata reviewed
Registry metadata indicates a reviewed listing.
Done
Safety and privacy checks
Complete
Validate risk disclosures before installation or API wiring.
Safety notes presentRequired
Review the listed safety guidance before running commands.
Done
Privacy notes presentRequired
Review data handling notes before connecting accounts or secrets.
Done
Trust level risk gateRequired
Trust level does not block evaluation.
Done
Package and install checks
Needs review
Check package metadata and artifact integrity signals.
Install payload available
Install or copy payload is available for review.
Done
Package verification flag
No package verification flag provided.
Pending
Checksum metadata
No checksum provided for downloaded artifact.
Pending
Compare-driven decision checks
Needs review
Use compare context to validate trade-offs before adoption.
Compare tray has multiple entries
Add at least one more entry to compare trust differences.
2 safety and 2 privacy notes across 3 risk areas. Review closely: credentials & tokens, network access.
3 areas
SafetyLocal filesPersists remembered information to a local file on disk (its location can be set with MEMORY_FILE_PATH); manage and review that file as it accumulates data.
SafetyNetwork accessRuns locally as an stdio process and makes no external network calls; the knowledge graph stays on your machine.
PrivacyCredentials & tokensDetails Claude chooses to remember are written to and read from the local memory file, so it can retain personal or project information across sessions.
PrivacyLocal filesReview or clear the memory file if it should not retain sensitive data, and use MEMORY_FILE_PATH to control where it is stored.
Safety notes
Persists remembered information to a local file on disk (its location can be set with MEMORY_FILE_PATH); manage and review that file as it accumulates data.
Runs locally as an stdio process and makes no external network calls; the knowledge graph stays on your machine.
Privacy notes
Details Claude chooses to remember are written to and read from the local memory file, so it can retain personal or project information across sessions.
Review or clear the memory file if it should not retain sensitive data, and use MEMORY_FILE_PATH to control where it is stored.
Prerequisites
Node.js 18+ and npx available (verify with: npx --version)
Claude Code or Claude Desktop with MCP support
A writable location for the memory file, optionally set with the MEMORY_FILE_PATH environment variable
Memory is one of the official Model Context Protocol reference servers. It gives Claude a persistent store that survives across sessions by maintaining a knowledge graph: entities (things worth remembering), relations (how they connect), and observations (facts attached to them). Instead of forgetting context when a conversation ends, Claude can record durable information and recall it later. Everything is kept in a local file on your own machine, so the memory stays under your control and the server makes no external calls.
Features
Persistent memory backed by a local knowledge graph of entities, relations, and observations.
Recall of stored information across separate Claude sessions.
Create, update, and query the graph as a normal part of conversation.
Storage in a local file whose path you can set with MEMORY_FILE_PATH.
Runs locally as a standard stdio MCP server with no external network calls.
Maintained as part of the official Model Context Protocol reference servers.
Use Cases
Remember durable project facts, preferences, or decisions between sessions.
Build up context about a codebase or domain over time.
Keep track of people, components, or systems and how they relate.
Give an assistant workflow continuity without re-explaining background each time.
Maintain a private, local memory store you fully control.
Installation
Claude Code
Make sure Node.js 18+ is installed (verify with npx --version).
Run: claude mcp add memory -- npx -y @modelcontextprotocol/server-memory
Verify the server is registered: claude mcp list
Ask Claude to remember a fact, then start a new session and ask it to recall.
Claude Desktop
Open your Claude Desktop configuration file.
Add the Memory server to the mcpServers section using the configuration below.
Optionally set MEMORY_FILE_PATH to choose where the memory file is stored.
Restart Claude Desktop and confirm the server appears.
To control where memory is stored, add an env block with MEMORY_FILE_PATH pointing to a writable file path.
Examples
Remember a durable fact
Ask Claude to store something for later.
"Remember that this project uses pnpm workspaces and deploys to Cloudflare."
Recall in a later session
Start fresh and ask Claude to use its memory.
"What do you remember about how this project is built and deployed?"
Relate entities
Build connections in the knowledge graph.
"Remember that the billing service depends on the auth service, and note why."
Security
The server writes remembered information to a local file. Treat that file as data storage: know where it lives, and back it up or clear it as appropriate.
It runs locally as an stdio process and does not send the knowledge graph to any external service.
Because it can retain personal or project details across sessions, review what is stored if the environment should not keep sensitive data.
Use MEMORY_FILE_PATH to place the memory file in a location with appropriate access controls.
Troubleshooting
npx cannot resolve the package
Verify Node.js 18+ and npx are installed (npx --version). Networks that block the npm registry will prevent npx -y @modelcontextprotocol/server-memory from resolving.
Memory does not persist between sessions
Confirm the process can write to the memory file location. If you set MEMORY_FILE_PATH, make sure the path is writable and consistent across runs so the same file is reused.
Server not listed in Claude Code
Re-run claude mcp add memory -- npx -y @modelcontextprotocol/server-memory, then claude mcp list. Confirm it was added to the scope (project versus user) you are running in.
Claude does not store or recall information
Prompt it explicitly, for example "remember this" or "use your memory". The model decides when to call the tools, so a clear instruction helps it write to or read from the graph.
Show that Memory MCP Server for Claude is listed on HeyClaude. Paste this Markdown into your README — it renders the badge and links back to this page.
[](https://heyclau.de/entry/mcp/memory-mcp-server)
How it compares
Memory MCP Server for Claude side by side with 3 alternatives on trust, install, platform support, and disclosed safety notes — all from reviewed registry metadata.
2 trust signals differ across this comparison (Source provenance, Submitter).
Official reference MCP server that gives Claude persistent memory through a local knowledge graph of entities, relations, and observations, stored in a file on your own machine.
ButlerBrain hosted MCP server providing persistent memory for AI assistants so Claude can store and recall context across sessions via api.butlerbrain.ai.
Local-first Basic Memory MCP server for giving Claude a persistent Markdown knowledge base with note editing, semantic search, recent activity, knowledge graph context, projects, schema tools, and optional cloud sync.
High-performance MCP server that indexes codebases into a persistent knowledge graph for structural search, call tracing, architecture summaries, dead-code detection, and cross-repo analysis.
✓Persists remembered information to a local file on disk (its location can be set with MEMORY_FILE_PATH); manage and review that file as it accumulates data.
Runs locally as an stdio process and makes no external network calls; the knowledge graph stays on your machine.
✓Persistent memory can retain incorrect or outdated facts until explicitly updated or deleted.
Memory write tools may store sensitive information if prompts include secrets or credentials.
Shared ButlerBrain workspaces can expose memories to other authorised users in the same account.
Review memory contents periodically to prevent unbounded accumulation of stale context.
✓Basic Memory MCP can create, edit, move, and delete notes in the configured project.
The server exposes content, search, knowledge graph, project, schema, cloud, and release-note tools.
Edit and delete tools should be reviewed carefully because the model can change the human-readable Markdown source of truth.
Imports can ingest Claude, ChatGPT, or memory JSON exports and may bring old sensitive conversations into the knowledge base.
Auto-update behavior can update uv tool and Homebrew installs, while uvx runs use the ephemeral runtime managed by uv.
The optional cloud service changes the trust boundary by syncing notes through hosted storage and account authentication.
✓Codebase Memory MCP reads source files across indexed repositories and writes MCP entries, instruction files, skills, hooks, and agent configuration files during install.
The background watcher and auto-index features can keep graph data updated as source files change.
The optional graph UI exposes a local visualization server on a localhost port.
Shared graph artifacts and SQLite databases can persist source metadata after an MCP session ends.
Verify downloaded release binaries, checksums, signatures, and provenance before running them in sensitive environments.
Privacy notes
✓Details Claude chooses to remember are written to and read from the local memory file, so it can retain personal or project information across sessions.
Review or clear the memory file if it should not retain sensitive data, and use MEMORY_FILE_PATH to control where it is stored.
✓Memories you save are stored on ButlerBrain infrastructure under its privacy and retention terms.
Do not store passwords, API keys, health records, or other regulated data in assistant memory.
Exported memory dumps may contain personal preferences; treat them as confidential user data.
✓Local notes, frontmatter, wikilinks, observations, relations, embeddings, logs, and project metadata can contain personal data, credentials, customer details, unreleased plans, or private research.
Cloud routing and hosted team workspaces may sync Markdown, metadata, snapshots, and backups outside the local machine.
The README says telemetry records non-content usage events and does not collect file contents, note titles, knowledge base data, PII, IP addresses, or per-tool tracking.
Logs are written to Basic Memory's local log file by default for MCP usage; review log level and retention before using sensitive prompts or notes.
Keep API keys, cloud credentials, imported chat archives, and private project paths out of prompts, screenshots, issues, and committed files.
✓Source code, filenames, paths, symbols, comments, imports, routes, call graphs, architecture summaries, ADRs, graph queries, prompts, and tool outputs may be visible to the MCP client and model provider.
Persistent graph databases and shared artifacts can retain proprietary architecture, internal package names, route names, service boundaries, and accidentally committed secrets.
Auto-indexing broad workspace roots can include unrelated private repositories or generated files.
Avoid committing shared graph artifacts unless the repository owners explicitly approve storing derived code intelligence in version control.
Prerequisites
Node.js 18+ and npx available (verify with: npx --version)
Claude Code or Claude Desktop with MCP support
A writable location for the memory file, optionally set with the MEMORY_FILE_PATH environment variable
ButlerBrain account with memory storage enabled for your workspace.
Claude Pro, Team, or Enterprise with Connectors support, or Claude Code with HTTP MCP transport.
Data retention policy review before storing customer PII or regulated data in persistent memory.
Understanding of what should and should not be remembered across sessions.
Python 3.12 or newer available through uv for the local MCP server.
A selected Basic Memory project directory for Markdown notes and generated metadata.
Backups or version control for important notes before enabling model-assisted writes.
Team agreement on whether notes stay local, sync through the optional cloud service, or route per project.
A local repository or workspace you are authorized to index.
Platform-compatible Codebase Memory MCP release binary verified before use.
Agent configuration write access reviewed before running the installer.
Sensitive paths, generated files, vendored code, secrets, and private repositories reviewed before indexing.
Install
claude mcp add memory -- npx -y @modelcontextprotocol/server-memory
claude mcp add --transport http butlerbrain https://api.butlerbrain.ai/mcp
claude mcp add basic-memory -- uvx basic-memory mcp