AI and memory servers store what an agent learns, retrieve from vector indexes, or call other models. Note where embeddings are computed and where memories are stored before you connect one. 3574 published MCP servers match this ranking. The highest trust scores right now are localmodels at 55, tcgdex at 55, myco-brain at 55. The order is the published trust score: registry record, license, maintenance, popularity, documentation, and security signals. Featured slots, when an editor sets them, are labeled and do not change the score. Best AI and memory MCP servers is not a paid placement.
Every server that matches this ranking, ordered by trust score.
Local-first memory for AI agents: markdown source of truth, budget-aware recall, write gate.
A true MCP server where the agent is the model and the API key. No API keys.
CPU task-specific context preparation and explicitly saved local project notes through MCP.
Token-efficient semantic compression for numerical data. 95%+ token reduction.
Encrypted memory for Claude, on a server you deploy to your own Cloudflare account.
Local-first shared memory for multiple AI agents that compounds with use; Markdown + git, MCP + CLI.
Self-hosted, uncensored AI agent: ask it tasks, search its long-term memory and your documents.
Detect and redact PII and secrets (API keys, tokens) in text, HTML or PDF before it reaches an LLM.
Research methods for AI assistants: study design guidance, R and Python code, and citations
Search and understand videos without embeddings: contact sheets plus Jev typed decisions.
Shared memory for your AI tools: top-5 recall 98.9% on answerable LongMemEval-S questions. Premium.
Local-first memory of how you write: style profile and real reply examples from your chat exports
Review-gated declarative long-term memory with bounded read-only MCP access.
Persistent memory, hybrid search and a goal graph for AI agents, over stdio or remote HTTP.
MCP server for DeepSeek web Search & Fetch: scored web search + page-to-Markdown fetch
Deterministic sub-0.2ms AST security circuit breaker and agent containment engine.
Send a prompt to ChatGPT anonymously and get the answer with its web sources, as JSON.
Lineage-gated agent memory tools on Postgres + pgvector, enforced by row-level security.
Run any Relay model (image, video, audio, LLM), poll tasks, upload files and search Relay's docs.
Run a golden set and gate LLM model swaps on accuracy, with latency and cost per request.
面向企业交付的飞书多维表格 MCP Server —— 让 Claude Code / Cursor 安全读写业务台账
MCP Neo4j Knowledge Graph Memory Server
by markusvankempen · Shell
MCP server demo — tickets, schema discovery, auth, and PII gating. stdio + HTTP.
by kythekitty · TypeScript
Fact-check any claim with live web sources. Ed25519-signed receipts. Catches hallucinations.