Skip to content
MCP server

Visual Memory MCP

Local visual cache for coding agents using perceptual hashing and CLIP search

Runs locallyStatus and history
Install
Pricing
Open source
Platforms
CLI, macOS, Windows
Runs
Locally
Version
1.4.2
Listing updated
1 day ago

About Visual Memory MCP

Visual Memory MCP is an open-source MCP server for knowledge and memory. It runs locally on your machine and is started by your AI client.

Visual Memory MCP stores screenshots of web and app interfaces on your machine so a coding agent does not have to re-send and re-interpret the same image each turn. It runs as a stdio server and CLI from the npm package @putervision/vision-memory-mcp, needs Node 18.18 or newer, and the README describes it as zero-infrastructure and local-first.

Lookup happens in tiers. An in-memory cache and dHash or aHash matching handle exact or near-duplicate screens in under about five milliseconds; local CLIP embeddings searched in LanceDB catch layouts with small CSS changes; accessibility-tree comparison guards against false hits; and a vision model is the last fallback. The README claims more than 95 percent of input token space can be recovered.

Other features include registering design mockups as baselines to verify layout drift, snapshot save and diff, WebM or MP4 ingestion into a state timeline, element grounding to selectors and coordinates, and trajectory export for fine-tuning. The README lists 15 tools and CLI commands such as doctor, audit, metrics and view.

Install

Run in a terminal

claude mcp add vision-memory-mcp -- npx -y @putervision/vision-memory-mcp

Add to claude_desktop_config.json

{
  "mcpServers": {
    "vision-memory-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@putervision/vision-memory-mcp"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "vision-memory-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@putervision/vision-memory-mcp"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "vision-memory-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@putervision/vision-memory-mcp"
      ]
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "vision-memory-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@putervision/vision-memory-mcp"
      ]
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "vision-memory-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@putervision/vision-memory-mcp"
      ]
    }
  }
}

Tools this server provides (3)

Taken from the project documentation; the running server may differ.

manage_visual_spec
Register and verify design baselines.
compare_states
Compare UI states for layout drift.
export_trajectories
Export UI state transition paths.

What it does

Tiered cache
Matches screens through memory, perceptual hash, CLIP and vision fallback.
Visual baselines
Registers mockups and flags layout drift against them.
Snapshots and diffs
Saves visual memory checkpoints and compares them.
Video ingestion
Turns WebM or MP4 recordings into searchable keyframe states.
Element grounding
Maps elements to selectors and coordinates.
CLI diagnostics
Offers doctor, audit, metrics and a graph viewer.

Good for

  • Cut vision token costs in a UI-testing agent
  • Catch visual regressions against a design mockup
  • Index end-to-end test recordings as searchable states
  • Check which screens an agent has already seen

Pricing

Open source

Free, open-source npm package

Prices change. Check visionmemorymcp.com before you buy.

Visual Memory MCP FAQ

Does it run locally?

Yes, the README describes it as local-first with no infrastructure.

What Node version is needed?

Node 18.18 or newer.

Can it work offline?

A run flag skips the CLIP model download for air-gapped use.

How many tools does it expose?

The README lists 15 core MCP tools.

Is Visual Memory MCP free?

Yes. Visual Memory MCP is open source, so you can use and self-host it for free.

How do I add Visual Memory MCP to Claude Code?

Run this command in your terminal: claude mcp add vision-memory-mcp -- npx -y @putervision/vision-memory-mcp

What are the best alternatives to Visual Memory MCP?

The closest alternatives to Visual Memory MCP are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.

Alternatives to Visual Memory MCP

  1. REA

    MCP server

    CLI and MCP server that lets coding agents reverse engineer apps and binaries

    Runs locally
    • Open source
    • 464K GitHub stars
    • Developer tools
  2. Codebase Memory

    MCP server

    Local MCP server that builds a code knowledge graph for AI coding agents

    Runs locally
    • Open source
    • 48.7K GitHub stars
    • Knowledge and memory
  3. GitHub

    MCP server

    Official MCP server connecting AI assistants to GitHub repos, issues and workflows

    Not checked yet
    • Remote
    • Open source
    • 34.5K GitHub stars
    • Developer tools
  4. MCP toolkit giving coding agents symbol-level code retrieval and editing

    Runs locally
    • Open source
    • 31.5K GitHub stars
    • Developer tools
  5. XERJ

    MCP server

    Local search engine for AI agents with cited answers over your code and docs

    Runs locally
    • Open source
    • 19.2K GitHub stars
    • Knowledge and memory
  6. Ouroboros

    MCP server

    Spec-first layer that pins requirements and verifies results of AI coding agent work

    Runs locally
    • Open source
    • 6.4K GitHub stars
    • Developer tools