"memory" matching MCP servers:

Provides persistent AI memory and visual analysis tools for any MCP-compatible agent via the iGenius Memory service.

Enables high-quality image generation using FLUX.1-dev through Claude Desktop or CLI, with automatic model unloading to save VRAM and memory-efficient bfloat16 processing.

An MCP server for local image, speech, music, and SFX generation that preserves character identity across calls and rejects degenerate outputs. It runs with near-zero GPU idle memory and is designed for DeepSeek Harness agents.

Enables AI agents to edit videos locally by inspecting media, applying templates, cutting clips, and delivering finished files, with memory of human preferences.

An MCP server for TouchDesigner that lets AI agents inspect, build, wire, optimize, and stabilize live TD networks with 106 tools, plus a technique memory system for reusable patterns.

Gives AI agents eyes on in-memory image buffers in live gdb/lldb C/C++ debug sessions: list observable symbols at a breakpoint, view renderings, read exact pixel values, and dump lossless .npy copies. Agents can also drive the human's viewer window (pan, zoom, channels, auto-contrast).

This MCP server aids users in searching and analyzing their photo library by location, labels, and people, offering functionalities like photo analysis and fuzzy matching for enhanced photo management.

Enables AI agents to act as graphic designers by generating images and social-ready visuals, with brand DNA memory, genre-aware styles, 35+ platform presets, self-critique, and version control.

An MCP server that orchestrates image generation across multiple pipelines (OpenAI, Recraft, Midjourney) with durable memory for project context, running on Cloudflare Workers.

Enables Codex sessions to observe and control the Windows desktop through screenshot reasoning, archived-frame inspection, pixel sampling, keyboard/mouse input, and workspace-local multimodal memory.

Provides a visual memory and attention layer for AI agents, enabling indexing of images and videos and then searching, inspecting, tracking, and retrieving the right frame in milliseconds via MCP tools.

A local stdio MCP server that reads a PDF page from a local path or authorized download URL and renders it in memory, returning a real PNG image block plus page metadata. This lets an agent's vision model actually see handwriting, formulas, arrows, colors, and figures instead of only extracted text, with configurable DPI and clipping while reporting page count and size.

Enables conversational image generation, editing, and refinement through OpenAI models with session memory for iterative creative workflows.

Enables AI visual debugging on Windows by capturing specific application, IDE, and compiled-app windows directly into memory, with background, high-DPI, side-by-side, and optional ShareX workflow support.

Enables AI agents to inspect and edit live Three.js scenes in the browser — objects, materials, shaders, textures, lights, animation, performance and memory — with 69 tools covering scene trees, screenshots, website audits and video frame capture. Changes can be previewed live with zero project modifications, then converted into ready-to-paste vanilla Three.js or React Three Fiber code.

Enables hands-free, context-aware control of a smart home by exposing MCP tools for device control, routine orchestration with confirmation flows, object finding, scene description, recent event history, and daily briefings. It keeps a mock home-state model with SQLite event memory and includes a web simulator that acts as an MCP client.

Enables local image analysis and document processing through MCP workers (Vision and Document) using Ollama models, with cloud orchestration for mixed hierarchical tasks. Supports tools for image understanding, requirement extraction, test case generation, and GPU memory lifecycle management.

Enables an MCP client to retrieve live World of Warcraft 3.3.5a character data — such as location, target, quests, bags, equipment, spells, stats, and group info — plus screenshots, addon listings, saved snapshots, and recent combat log lines. It does this read-only by pairing an in-game addon's color-strip output with local screen capture and decoding, without reading game memory, injecting code, or controlling the character.

Enables persistent out-of-model memory plus screen, camera, browser, video, and messaging tools through MCP, with per-capability permissions, token cost tracking, and fail-closed approvals. Supports recalling preferences, viewing the screen or camera, collecting and saving video content, drafting or sending messages, switching character appearances, and using external models across MCP clients.

Gives an Alexa+ voice agent a household memory layer that enforces allergy rules, grounds answers to a child's reading level, and checks existing ports and spare cables before any purchase. It runs as a self-hosted Streamable HTTP MCP server with eight tools, including Bedrock-powered multimodal inspection of unboxed deliveries.