@putervision/vision-memory-mcpCAUTION
Persistent visual cache for LLM-driven software development. Caches screenshots using perceptual hashing, vector search, and AX trees to prevent token overhead and visual hallucination loops.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
[](https://www.npmjs.com/package/@putervision/vision-memory-mcp) [](./CHANGELOG.md) [](https://www.npmjs.com/package/@putervision/vision-memory-mcp) [](https://github.com/putervision/vision-memory-mcp/actions/workflows/ci.yml) [](https://nodejs.org) [](https://www.typescriptlang.org) [](https://visionmemorymcp.com) [](https://github.com/putervision/vision-memory-mcp/blob/main/LICENSE)
@putervision/vision-memory-mcp is a zero-infrastructure, local-first Model Context Protocol (MCP) server and CLI tool that provides AI coding assistants (such as Cursor, Claude Code, Gemini, or Copilot) with visual state caching using perceptual hashing, local CLIP embeddings, and transition graphs to eliminate repetitive vision LLM calls.
🌐 Official Documentation & Website: visionmemorymcp.com
⚡ Quick Start & Installation
Prerequisites: Node.js >= 18.18.0
1. Installation
# Global installation via npm npm install -g @putervision/vision-memory-mcp
2. Workspace Initialization
Run init in your project root to scaffold database directories, .gitignore, .env, and IDE rules:
vision-memory-mcp init --yes
3. Basic MCP Client Setup
Add to your MCP client config (e.g. .cursor/mcp.json or .vscode/mcp.json):
aaa65c673d26OBSERVED · 2026-10-12Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add vision-memory-mcp -- npx -y @putervision/[email protected]
Exposed tools (30)
28 read · 1 write · 1 destructive. Blast radius: 1 tool can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.
| Tool | Risk | Description |
|---|---|---|
analyze_screenshot | read | |
checkpoint-alpha | read | First checkpoint |
checkpoint-beta | read | Second checkpoint |
compare_states | read | |
create_evidence_pack | write | |
custom_content_tool | read | |
delete-me-snap | destructive | Snapshot description |
dummy | read | dummy |
echo_tool | read | |
export_trajectories | read | |
failing_tool | read | |
forget_state | read | |
get_navigation_paths | read | |
get_session_context | read | |
manage_snapshot | read | |
manage_video | read | |
manage_visual_spec | read | |
predict_next_action | read | |
recall_memory | read | |
record_outcome | read | |
snap-for-export | read | Export test |
snapshot-b | read | After additions and drift |
string_tool | read | |
test-snap-1 | read | Checkpoint 1 |
test-snap-2 | read | Checkpoint 2 |
test_tool | read | |
throw_tool | read | |
tool-with-zod | read | |
undo_visual_mutation | read | |
wait_for_visual_state | read |
Trust audit
CAUTIONgrade B · trust 80/100 Install with care. The audit found things worth knowing before you trust its output.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | WARN |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | WARN |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (7 observation(s))
- Network
- declared (1 observation(s))
- Shell
- declared (2 observation(s))
- Dependencies
- not all pinned
- Secrets in source
- found
Findings (16)
api_key: 'sk-abcdef123456789012345678',
Token: ghp_123456789012345678901234567890123456
'API key sk-proj-1234567890abcdef1234567890abcdef and token ghp_123456789012345678901234567890123456';
delete-me-snap
.prettierignore
.windsurfrules
await expect(resolveImageInput(undefined, '/home/user/.ssh/id_rsa')).rejects.toThrow(
import { storage, escapeSql } from '../../core/storage.js';import { getCurrentBranch, memoryCache } from '../../core/cache.js';import { config } from '../../config.js';import { logger } from '../../logger.js';const { repairStorageHealth } = await import('../../core/storage-health.js');targetUrl: 'http://127.0.0.1:9999',
'iVBORw0KGgoAAAANSUhEUgAAAEAAAABACAMAAACdt4HsAAAAAXNSR0IArs4c6QAAAARnQU1BAACxjwv8YQUAAAADUExURQAAAP851ZcAAAAZdEVYdFNvZnR3YXJlAEFkb2JlIEltYWdlUmVhZHlxyWU8AAAAE0lEQVRYR+3BAQ0AAADCoPdPbQ8HFAAAuBsCNAABz4r
'iVBORw0KGgoAAAANSUhEUgAAAEAAAABACAMAAACdt4HsAAAAAXNSR0IArs4c6QAAAARnQU1BAACxjwv8YQUAAAADUExURQAAAP851ZcAAAAZdEVYdFNvZnR3YXJlAEFkb2JlIEltYWdlUmVhZHlxyWU8AAAAE0lEQVRYR+3BAQ0AAADCoPdPbQ8HFAAAuBsCNAABz4r
@huggingface/transformers, @eslint/js, @types/node, @vitest/coverage-v8, eslint, prettier, tsup, typescript
Gates applied: no_behavioural_pass.
aaa65c673d26full audit observations/trust-audit/mcp-server/putervision__vision-memory-mcp.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-12 | aaa65c673d26 | CAUTION | B | 80 | first audit |
Questions
What is the @putervision/vision-memory-mcp MCP server?
Persistent visual cache for LLM-driven software development. Caches screenshots using perceptual hashing, vector search, and AX trees to prevent token overhead and visual hallucination loops.
What tools does @putervision/vision-memory-mcp expose?
30 in total: 28 read-only, 1 that write, and 1 that can delete or overwrite (delete-me-snap). Every one is listed on this page with its risk.
Is @putervision/vision-memory-mcp safe to connect to an agent?
With care. The audit graded it B (80/100) and found 16 things worth knowing before you trust this server, listed below with the exact line each was found on. Separately from the audit: 1 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does @putervision/vision-memory-mcp need?
It reads OPENAI_API_KEY from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does @putervision/vision-memory-mcp run?
It speaks stdio, so it runs as a local process your client starts. It is published on npm as @putervision/vision-memory-mcp at 1.4.2.
How current is this page?
The grade is for one exact copy of the source (aaa65c673d26), read on 2026-10-12. The repository is watched and re-audited when it changes.