# MemoFS MCP Server (npm · @memofs/mcp-server)

Persistent memory and virtual filesystem MCP server for AI agents.

- Trust score: 95/100 (high trust)
- Change this week: 0
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-09-20

## Components

- remote · `memofs.dev`: 23/100, [markdown](https://verifymcp.io/servers/dev-memofs-mcp-server/api-v1-projects-projectid-mcp.md), [page](https://verifymcp.io/servers/dev-memofs-mcp-server/api-v1-projects-projectid-mcp)
- npm · `@memofs/mcp-server`: 95/100 (this document), [markdown](https://verifymcp.io/servers/dev-memofs-mcp-server/memofs-mcp-server.md), [page](https://verifymcp.io/servers/dev-memofs-mcp-server/memofs-mcp-server)

## Channel facts

- Registry: `npm`
- Package: `@memofs/mcp-server`
- Version: `1.3.0-beta.3`
- Transport: `stdio`

## Trust breakdown

How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog)

Scored 2026-09-20.

- **Supply Chain Security**: 98/100
  - No malware found by supply-chain analysis.
  - No known CVEs affecting this package version or its production dependencies.
  - No install/post-install scripts declared.
  - 31 of 97 dependencies flagged as unhealthy.
- **Provenance & Transparency**: 100/100
  - Source repository is publicly reachable at the declared URL.
  - Cryptographically verified build provenance (signed, bound to memo-fs/memofs).
  - Clear OSI-approved license (MIT).
  - Actively maintained (last published 33 days ago).
  - Publishes a security disclosure policy (SECURITY.md).
- **Schema Quality & AI Usability**: 84/100
  - 100% of prompts and resources have a non-trivial description (not blank, and not just the item's name).
  - AI-judged instruction clarity (excellent).
  - Context-footprint check failed: tool/resource definitions use about 2157 tokens (~113/item across 19 items; 10 tools + 9 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 90/100
  - Stability observed for 27 of 30 days with no destabilising changes; credit accrues until the full window elapses.
- **Tool Coverage**: 99/100
  - 100% of tools have a non-trivial description (not blank, and not just the tool's name).
  - 97% of tool parameters carry a description.
- **Tool Safety**: 100/100
  - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.
  - We read all 10 captured tool definition(s), and no name or description among them implies an irreversible operation.
  - An AI judge read all 12 captured unit(s) of tool text and found none that tries to manipulate the model reading it.
- **Capabilities**: 100/100
  - Implements a supported MCP spec version (2025-11-25); the latest is 2026-07-28.

## Install

### How do I install the MemoFS MCP Server server?

MemoFS MCP Server runs locally as an npm package, launched with npx -y @memofs/mcp-server. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.

### Claude

```bash
claude mcp add dev-memofs-mcp-server -- npx -y @memofs/mcp-server
```

### Cursor

```json
{
  "mcpServers": {
    "dev-memofs-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "@memofs/mcp-server"
      ]
    }
  }
}
```

### VS Code

```json
{
  "servers": {
    "dev-memofs-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "@memofs/mcp-server"
      ]
    }
  }
}
```

### Codex

```bash
codex mcp add dev-memofs-mcp-server -- npx -y @memofs/mcp-server
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "dev-memofs-mcp-server": {
      "type": "local",
      "command": [
        "npx",
        "-y",
        "@memofs/mcp-server"
      ],
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add dev-memofs-mcp-server --command npx --arg -y --arg @memofs/mcp-server
```

### Hermes

```yaml
mcp_servers:
  dev-memofs-mcp-server:
    command: "npx"
    args: ["-y", "@memofs/mcp-server"]
```

### Netclaw

```json
{
  "McpServers": {
    "dev-memofs-mcp-server": {
      "Transport": "stdio",
      "Command": "npx",
      "Arguments": [
        "-y",
        "@memofs/mcp-server"
      ]
    }
  }
}
```

### Vellum

```bash
assistant mcp add dev-memofs-mcp-server -t stdio -c npx -a -y @memofs/mcp-server
```

### Other

```json
{
  "mcpServers": {
    "dev-memofs-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "@memofs/mcp-server"
      ]
    }
  }
}
```

## Changelog

Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims.

### 2026-09-19 (score 95, +1)

No change was recorded against any check on this day. Stability & Change Management went from 83 to 87. That category is still filling its 30-day observation window: 25 days of observed history at the previous scan, 26 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-17 (score 94, −2)

- [functional] Stability: pass → 0.80

### 2026-09-16 (score 96, 0)

- [security] Stability: 0.97 → pass

### 2026-09-15 (score 96, +1)

No change was recorded against any check on this day. Stability & Change Management went from 93 to 97. That category is still filling its 30-day observation window: 28 days of observed history at the previous scan, 29 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-12 (score 95, +1)

No change was recorded against any check on this day. Stability & Change Management went from 83 to 87. That category is still filling its 30-day observation window: 25 days of observed history at the previous scan, 26 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-10 (score 94, +1)

No change was recorded against any check on this day. Stability & Change Management went from 77 to 80. That category is still filling its 30-day observation window: 23 days of observed history at the previous scan, 24 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-08 (score 93, +1)

No change was recorded against any check on this day. Stability & Change Management went from 70 to 73. That category is still filling its 30-day observation window: 21 days of observed history at the previous scan, 22 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-06 (score 92, +1)

No change was recorded against any check on this day. Stability & Change Management went from 63 to 67. That category is still filling its 30-day observation window: 19 days of observed history at the previous scan, 20 at this one. The score rises as the window fills, whether or not the server changes.

## MCP tools (10)

### `memofs_agent_session_start` (~98 tokens)

Start MemoFS Agent Session

Create an AgentFS-backed MemoFS session workspace and return paths/resources for coding agents.

Input parameters:

- `actorId` (string): Actor id such as assistant:codex.
- `projectId` (string): Project id inside the workspace.
- `sessionId` (string): Optional safe session id.
- `task` (string, required): Agent task or brief.
- `workspaceId` (string): Workspace id. The runtime decides whether this is required.

### `memofs_agent_session_read` (~69 tokens)

Read MemoFS Agent Session File

Read one AgentFS session file.

Input parameters:

- `path` (string, required): Session file path.
- `projectId` (string): Project id inside the workspace.
- `sessionId` (string, required): Session id.
- `workspaceId` (string): Workspace id. The runtime decides whether this is required.

### `memofs_agent_session_write` (~86 tokens)

Write MemoFS Agent Session File

Write an allowed working/ or output/ file in an AgentFS session.

Input parameters:

- `content` (string, required): File content.
- `path` (string, required): Session file path.
- `projectId` (string): Project id inside the workspace.
- `sessionId` (string, required): Session id.
- `workspaceId` (string): Workspace id. The runtime decides whether this is required.

### `memofs_agent_session_append` (~88 tokens)

Append MemoFS Agent Session File

Append to an allowed working/ or output/ file in an AgentFS session.

Input parameters:

- `content` (string, required): Content to append.
- `path` (string, required): Session file path.
- `projectId` (string): Project id inside the workspace.
- `sessionId` (string, required): Session id.
- `workspaceId` (string): Workspace id. The runtime decides whether this is required.

### `memofs_agent_session_extract` (~71 tokens)

Extract MemoFS Agent Session Memory

Extract summary, durable memory, follow-ups, errors, and changes from an AgentFS session.

Input parameters:

- `projectId` (string): Project id inside the workspace.
- `sessionId` (string, required): Session id.
- `workspaceId` (string): Workspace id. The runtime decides whether this is required.

### `memofs_agent_session_complete` (~359 tokens)

Complete MemoFS Agent Session

Extract, checkpoint, push, and optionally persist durable memory from an AgentFS session. Set `outcome` explicitly — the `"success"` default is for backward-compatibility only and is NOT recommended for new callers. `outcome: "success"` + `extractDurableMemory: true` promotes durable memory to notes.md and auto-cleans `working/` (output preserved). `outcome: "failure"` never promotes durable; pass `ephemeral: true` to clean `working/` + `output/`, or omit to preserve them as failure audit-trail. `outcome: "aborted"` preserves the entire workspace for resume via a future `complete({outcome:"success"|"failure"})` with the same `sessionId`. A `reason` string on `failure`/`aborted` is recorded in a `session.failed` audit event for telemetry.

Input parameters:

- `checkpointLabel` (string): Checkpoint label.
- `ephemeral` (boolean): On outcome: "failure", opt-in cleanup of working/ + output/ session files. Ignored for other outcomes.
- `extractDurableMemory` (boolean): Persist output/durable-memory.md into MemoFS notes (first gate; requires outcome: "success" to promote).
- `outcome` (string): Session outcome. Defaults to "success" for backward-compatibility. Set explicitly — new callers should NOT rely on the default.
- `projectId` (string): Project id inside the workspace.
- `reason` (string): Structured failure/abort audit text; carried on the session.failed audit event.
- `sessionId` (string, required): Session id.
- `workspaceId` (string): Workspace id. The runtime decides whether this is required.

### `memofs.context` (~466 tokens)

Build MemoFS Agent Context

REQUIRED at the start of every task: build task-ready memory context by combining core memory, recent memory, and recall. MemoFS is the single source of truth for project identity, architecture, constraints, and decisions — ALWAYS call this before planning or writing code so your work adheres to stored memory. The returned text already tells you how to act on it. Read-only.

By default returns a COMPACT briefing (~6kb): core memory in full, the top entities and a few recall fragments, plus an `expandable` list. Each expandable entry names a section (entities/recall/recent/notes), how many more items it holds, and an opaque cursor. To pull just one section in full, call memofs.context again with the same query plus `section` and `expand` (the cursor). Expand only the sections you need, then stop — do not expand everything. Pass `detail: "full"` to get the entire context in one call instead (larger; use when you want the whole dump).

Input parameters:

- `detail` (string): Disclosure level. "compact" (default): a small briefing with expandable sections. "full": the entire context in one call.
- `expand` (string): Opaque expansion cursor returned by a prior compact call. Set together with `section`.
- `filters` (object): Runtime-specific JSON filters.
- `includeCore` (boolean): Include core memory.
- `includeGraph` (boolean): Ask runtime for graph-aware context.
- `includeNotes` (boolean): Include notes memory.
- `includeRecent` (boolean): Include recent memory events.
- `includeSources` (boolean): Include source references.
- `limit` (integer): Maximum recall items to include.
- `maxBytes` (integer): Maximum context text bytes.
- `projectId` (string): Project id inside the workspace.
- `query` (string, required): Current user task or context query.
- `section` (string): Expand a single section: set together with `expand` to the cursor a compact call returned.
- `taskType` (string): Kind of task the agent is performing. The strategist tailors the recall query per task type. Defaults to "general".
- `workspaceId` (string): Workspace id. The runtime decides whether this is required.

### `memofs.recall` (~207 tokens)

Recall MemoFS Memory

Semantic + lexical memory search. Use this proactively — before answering, when unsure, or when a fact might already be known. Phrases it understands: synonyms and paraphrases, not just exact keywords. Call it instead of guessing or re-deriving facts. Read-only; never modifies memory. Each returned item may carry an optional `anchor` (the code file + hash this fact is bound to) and a `stale` flag (true when the anchored file changed since write — re-verify before trusting).

Input parameters:

- `filters` (object): Runtime-specific JSON filters.
- `includeGraph` (boolean): Whether to include graph-aware recall if supported.
- `includeSources` (boolean): Whether to include source references.
- `limit` (integer): Maximum results to return.
- `projectId` (string): Project id inside the workspace.
- `query` (string, required): Natural language recall query.
- `workspaceId` (string): Workspace id. The runtime decides whether this is required.

### `memofs.remember` (~237 tokens)

Remember MemoFS Memory

Persist a durable fact so future agents benefit — call this WITHOUT being asked whenever you discover a decision, constraint, preference, or architectural fact. Use kind to classify (decision/constraint/goal/preference/reference/summary/note). Set confidence to reflect certainty. Hosts may require user consent; never store secrets. This is what makes memory accumulate intelligently.

Input parameters:

- `anchor` (object): Code anchor: bind this memory to a file's bytes for drift detection. When set, recall recomputes the file's SHA-256 and flags this memory stale if the bytes changed or the file was deleted.
- `confidence` (number): Confidence score.
- `content` (string, required): Memory note content.
- `kind` (string): Memory kind. Defaults to note.
- `metadata` (object): JSON metadata.
- `projectId` (string): Project id inside the workspace.
- `source` (string): Source actor or origin label.
- `sourceRefs` (array)
- `tags` (array)
- `title` (string): Optional title.
- `workspaceId` (string): Workspace id. The runtime decides whether this is required.

### `memofs.consolidate` (~201 tokens)

Consolidate MemoFS Graph Memory

Run a memory consolidation pass over the local graph: merge duplicate entities and retire facts superseded by a `supersedes` edge. The audit trail is preserved — nothing is deleted, only marked `deprecated`. This is the second half of v1 intelligence: extraction grows the graph, consolidation keeps it tidy. Pass apply=false to preview the plan without persisting. Hosts should authorize this write when apply is true.

Input parameters:

- `apply` (boolean): Persist the computed plan (merges + retirements). Defaults to true; pass false to preview.
- `now` (string): Override the `now` timestamp stamped on retirements (ISO 8601). Mainly for tests.
- `projectId` (string): Project id inside the workspace.
- `supersedingEdgeType` (string): Edge type expressing 'A replaces B'. Defaults to 'supersedes'.
- `workspaceId` (string): Workspace id. The runtime decides whether this is required.

## Diagnostics

Captured diagnostic sections: Provenance, Dependencies. The full working is on the page: https://verifymcp.io/servers/dev-memofs-mcp-server/memofs-mcp-server#diagnostics

## Score history

- 2026-09-20: 95
- 2026-09-19: 95
- 2026-09-18: 94
- 2026-09-17: 94
- 2026-09-16: 96
- 2026-09-15: 96
- 2026-09-14: 95
- 2026-09-13: 95
- 2026-09-12: 95
- 2026-09-11: 94
- 2026-09-10: 94
- 2026-09-09: 93
- 2026-09-08: 93
- 2026-09-07: 92
- 2026-09-06: 92
- 2026-09-05: 91
- 2026-09-04: 91
- 2026-09-03: 91
- 2026-09-02: 90
- 2026-09-01: 90
- 2026-08-31: 89
- 2026-08-30: 89
- 2026-08-29: 88
- 2026-08-28: 88
- 2026-08-27: 87
- 2026-08-26: 87
- 2026-08-25: 85
- 2026-08-24: 81
- 2026-08-23: 81
- 2026-08-22: 81

## Common questions

### What is the MemoFS MCP Server server?

MemoFS MCP Server is listed in the public MCP registry as dev.memofs/mcp-server. Persistent memory and virtual filesystem MCP server for AI agents. This page covers its npm package (@memofs/mcp-server).

### Is the MemoFS MCP Server server safe to use?

MemoFS MCP Server scores 95 out of 100 on VerifyMCP. We found no known CVEs affecting it as of 20 September 2026. It declares no install or post-install scripts. Its build provenance is signed and verified. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

### What tools does the MemoFS MCP Server server expose?

MemoFS MCP Server exposes 10 tools: memofs_agent_session_start, memofs_agent_session_read, memofs_agent_session_write, memofs_agent_session_append, memofs_agent_session_extract, and 5 more. Their descriptions and schemas cost roughly 1,882 tokens of context every time the server is loaded.

### Is the MemoFS MCP Server server still maintained?

MemoFS MCP Server is still listed as active in the MCP registry. We last reached this channel on 20 September 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

### What licence is the MemoFS MCP Server server under?

MemoFS MCP Server declares the MIT licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls.

## Links

- npm package: https://www.npmjs.com/package/@memofs/mcp-server
- Socket report: https://socket.dev/npm/package/@memofs/mcp-server
- Repository: https://github.com/memo-fs/memofs
- Website: https://memofs.dev/
- Changelog RSS feed: https://verifymcp.io/servers/dev-memofs-mcp-server/memofs-mcp-server.xml
- Changelog JSON feed: https://verifymcp.io/servers/dev-memofs-mcp-server/memofs-mcp-server.json
- HTML version of this page: https://verifymcp.io/servers/dev-memofs-mcp-server/memofs-mcp-server