agentmods.dev

vision-memory-mcp

A set of rules for using visual memory to cache screen descriptions, track layout changes, identify interface elements, and avoid repeating visual analysis. It is built around vision-memory-mcp.

In plain words
What is it for?
Use it when inspecting screenshots, navigating changing interfaces, grounding actions to selectors or coordinates, recording transitions, and removing stored visual states for privacy.
Why use it?
It keeps known screen states and reliable selectors available, reducing repeated image analysis and making clicks or typing more consistent.

Cursor rule for Cursor

26 10d ago B scan 875 tokens copy · 100% MIT
View the source file putervision/WebCrypt
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add rules/putervision/webcrypt/vision-memory-mcp
Clone the repo
git clone --depth 1 https://github.com/putervision/WebCrypt

Made for: Cursor.

Per session 875 This file is loaded in full into every session.
When invoked 875 The same file — it is already loaded in full.
Security scan B 1 finding. Scan, not verified.
Origin 100% copy Near-identical to another mod in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00875 $0.00875
Opus 5 $0.00438 $0.00438
Sonnet 5 $0.00175 $0.00175
Haiku 4.5 $0.00088 $0.00088

Measured yesterday against content hash 64a612948bea, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

vision-memory-mcp scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

* **Google Antigravity (`~/.gemini/config/config.json`)**: Add these rules to your `"globalPermissionGrants"` -> `"allow"` list:
Origin

This is a copy

100% identical to vision-memory-mcp — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.

.cursor/rules/vision-memory-mcp.mdc · 42 lines

How it starts

The opening of the file, as written. The whole thing — 42 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Visual Memory (vision-memory-mcp)

This project utilizes vision-memory-mcp to cache visual states, record layout transitions, provide element grounding, and avoid repetitive LLM vision calls.

1. Mandatory Workflow & Priority

  1. Orient: Call get_session_context to align your visual state context at the start of work.
  2. Search: Call recall_memory (text/image search) before recreating duplicate UI state paths.
  3. Ingest/Verify: ALWAYS call analyze_screenshot before querying any front-end vision models.
    • Cache Hit (is_known: true): Do NOT use vision models; read the returned description as context and use grounded_elements (selectors, coordinates) for action target selection.
    • Cache Miss (is_known: false): Query your vision model, then run analyze_screenshot with both the image and description to seed the cache.
  4. Action Target Execution: Use predict_next_action to retrieve grounded_target handles (target_selector, target_coords) for deterministic UI clicks and typing.
  5. Transitions: Call record_outcome after every click/type/scroll action to construct navigation paths.
  6. Privacy & Cleanup: Call forget_state to purge sensitive or secret states from storage.

2. Tool Reference Summary (15 Core MCP Tools)

  • analyze_screenshot: Ingest screenshot(s) (single or batch via items), lookup cache, return layout description and grounded elements.
  • recall_memory: Search visual memory by description query or base64 image query (read-only).
  • record_outcome: Save UI action execution outcomes, transitions, or log visual blockers (action_type: 'blocker').
  • get_navigation_paths: Find path between states using BFS navigation graph.
  • predict_next_action: Predict best next UI action and target coordinates based on transition success rates and AX tree grounding.
  • compare_states: Compare visual states structurally (has_layout_change) or compare video recordings (video_a_id/video_b_id).
  • get_session_context: Fetch aggregated visual context, recent/frequent states, transitions, cache hit ratios, token savings metrics, and server version info.
  • manage_snapshot: Unified snapshot management (save, diff, export, restore) for visual checkpoints and regression detection.
  • manage_visual_spec: Visual SDD design contract baseline registration (set), live verification (verify), and listing (list).
  • manage_video: Unified video memory operations for ingestion (ingest), semantic search (search), and keyframe timelines (timeline).
  • create_evidence_pack: Create cryptographic, multi-modal evidence pack linking video keyframes, state graph tasks, and visual proof.
  • export_trajectories: Export multimodal visual transitions and joint workflow trajectories (json, llava, qwen2_vl, joint).
  • undo_visual_mutation: Revert accidental state or transition edge ingestions.
  • forget_state: Purge a specific state and vector embedding from storage for privacy.
  • wait_for_visual_state: Poll for target visual state until present or timeout occurs.

Read the full file on GitHub · 42 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 42 lines · 875 tokens per session scan B 64a612948bea

Subscribe to this mod's changes

vision-memory-mcp is a cursor rule published in the GitHub repository putervision/WebCrypt (26 stars, last pushed 10d ago), licensed MIT. It adds 875 tokens to every session, about $0.0044 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). It is 100% identical to vision-memory-mcp, differing in 0 lines, and is treated as a copy.

Data and AIMemory and contextSecuritycryptographye2eee2ee-encryptionencryptionhmacjwe
Related

Other cursor rules, from other repositories

angular-20

This rule provides comprehensive best practices and coding standards for Angular development, focusing on modern TypeScript, standalone components, signals, and performance optimizations.

angular/angular · 0 tokens

dev-standard

Apache Superset development standards and guidelines for Cursor IDE.

apache/superset · 1,147 tokens

typescript

Changes to these high-fan-out internals can affect every message, delta, element, or rerun. Keep work in them minimal, and benchmark changes with representative stress-test apps.

streamlit/streamlit · 4,669 tokens

coolify-ai-docs

Master reference to all Coolify AI documentation in .ai/ directory.

coollabsio/coolify · 1,426 tokens

python_lib

Tips and guidelines specific to the development of the Streamlit Python library, not applicable to scripts and e2e tests.

streamlit/streamlit · 0 tokens

specs

This directory contains product and tech specs for Streamlit features.

streamlit/streamlit · 0 tokens