Memory MCP Servers — Page 2

When an agent greps a codebase it either dumps whole files into context or misses the code entirely because it guessed different words. repo2graph builds the real call graph instead and uses it to decide what to return — so you get the actual source, inside a fixed token budget, with the callers and callees already attached.

A personal MCP context layer so you stop re-explaining data definitions to your coding agent every session. Connect Claude Code, Cursor, or Codex in one command. Free to start.

Memory for facts that change. A retrieved record is marked SUPERSEDED, with when a later record replaced it, so an assistant does not answer with a value that is no longer true.

The log your next session resumes from: decisions, dead ends, handoffs and open tasks, shared across machines, agents (Claude Code, Codex, Cursor, VS Code) and people. Stale notes are flagged, never faked; reports come from the log. Hosted or self-hosted, MIT.

One memory shared by Claude Code, Codex, Cursor, Copilot CLI, OpenClaw and 28 more coding agents, built from the session history already on disk. A fix found in one agent comes back in any of them, including months of sessions from before you installed it. No LLM, no embeddings, one local Go binary.

Read-only personal memory server: lets AI agents read a user's ORANO library — saved-video projects with summaries, key points, ordered tasks, research, and semantic search. Official registry: io.github.manuvamp/orano-personal-context-mcp

Hosted MCP memory for coding agents: persistent across sessions, editable markdown, team sharing.

Reasoning scaffolding that supercharges your small AI: an argument-mapping MCP server

A Support Center for Autonomous Artificial Minds

Prevents AI coding agents (Cursor, Claude Code) from losing context across 1,000+ turns and slashes LLM tokens by 96% in 75KB pure Rust.

把团队的重复 AI 工作铸成可治理的工作流资产：AI 写操作强制人审（staged writes）、一句话铸造工作流模板（评测守门）、三记忆推衍引擎主动建议工作流。

Local-first long-term memory for AI coding agents. Gives Claude Code, Cursor, Codex and other MCP-compatible agents persistent memory across sessions - project context, architecture decisions and past failures. Memory data stays on your own machine.

Verifiable memory for AI agents. MCP server with 4 memory planes (working, episodic, semantic, claims). Writes can be anchored on Starknet mainnet for independent, third-party verification. Free tier: 1,000 requests/day, no time limit.

Turn your Instagram saved posts into a searchable personal knowledge library — local SQLite, read-only MCP server

Local-first zero-LLM memory for AI agents with BM25, dense-vector, and reciprocal-rank-fusion retrieval.

Files what you learn from ChatGPT or Claude into a personal wiki, then schedules spaced-repetition review so you actually remember it. Your assistant can read and update the pages directly.

Personal context for every AI: search, read, and write back to your private Markdown library of articles, threads, PDFs, notes, and captured ChatGPT/Claude/Gemini/Grok conversations. Remote server, OAuth 2.1 paste-and-authorize or revocable tiered Agent keys (read_only / edit / full). Every agent edit is versioned and revertible.

Local temporal knowledge graph with six read-only MCP tools, Ollama inference, embedded LadybugDB, and a separate human-approved CLI write queue.

Streamable-HTTP MCP server for Synap long-term agent memory. Store, recall and forget tools with per-tenant scoping, backed by a managed memory service.

Self-improving skill memory for coding agents: skills that help get reinforced, unused ones decay on an Ebbinghaus curve. $0 write path (SQLite FTS5 + local ONNX embeddings, no LLM calls), bilingual EN/RU hybrid search, SHA256 tamper-evident history, reproducible LongMemEval benchmark (hit@5 0.871). 8 mem_* tools + optional Claude Code hooks.

Seeds, governs and navigates a memory_bank knowledge base: gated writes, routing without reading, and drift against the code

Persistent memory for AI agents — EU-hosted and privacy-first. Tools: remember, recall (hybrid semantic + lexical), supersede with audit chains, history, bootstrap. Write-time contradiction detection flags when a new fact clashes with stored knowledge. Works with Claude Code, Cursor, Zed and any MCP client. Free tier, no card — and it also runs fully local with no account at all.

Turns your LinkedIn/GitHub/course history into a queryable knowledge graph over MCP, using real semantic search instead of keyword matching.

Persistent memory layer any model can read and write — knowledge graph, LIFE tasks, CRM contacts and decisions that survive across Claude, GPT, Gemini and 50+ models.

Chat forgets your workouts. AIm remembers them for Claude and ChatGPT: sets, weights, 1RM, volume.

Hosted memory MCP server for saving and retrieving durable facts, preferences and decisions from compatible AI assistants. Supports authenticated memory tools and record inspection.

I chose memory for the catagory but mem20mpc has capabilities stemmimng from every one of the catagories available. Memory seemed most appropriate however. Because one of the flagship features is purely memory based Cross Session ToT evolutionary learnig system

Local-first persistent memory for MCP-compatible AI agents.

Evidence-chain memory for AI agents: verbatim-first (immutable SHA-256 hashed captures), auditable, anti-fabrication. Local-first skill + MCP server for Claude Code, Codex and any MCP client.

A local memory engine any AI tool can use.