知识与记忆 MCP 服务器

FAISS, call graph, AST, BM25 — 34 MCP tools for AI agents. 50-80% token reduction. Offline.

Persistent memory for AI coding tools via MCP. Remembers across sessions and machines.

Reasoning via a DAG thought graph with multiple strategies (dialectic, analogical, abductive and more), stuck detection, confidence…

Git-native project memory for AI coding agents: Markdown source of truth, reviewable, secret-safe.

Local-first memory for AI coding agents: shell history with exit codes, git diffs, docs, MCP.

Local long-term memory shared by all your AI agents, kept word for word on your own PC.

Two-tier memory: hot cache (0ms) + semantic search. Self-organizing.

Obsidian MCP server: semantic search, knowledge graph, and vault editing. No plugin required.

Hebbian memory for AI agents. Learns file patterns, predicts next tools, saves tokens.

Local-first, source-traceable agent memory — no LLM at ingest, fully offline

Local memory for Claude Code: 13 MCP tools for semantic recall, file replay, and git history.

Search Lenny's Podcast transcripts with YouTube timestamps for product management insights on PRDs, strategy and PM careers.

Local-first, source-cited memory for AI agents with proposal review, failure memory, and MCP tools.

Search Google's Knowledge Graph for structured information about real-world entities

Persistent memory for AI assistants: store, search, and connect knowledge across conversations.

Structured, persona-driven sessions for interview preparation, personal reflection and coaching, with timers and performance evaluation.

Access Obsidian vaults via Local REST API - read, search, and interact with notes

Five-level AI collaboration system with persistent local-first memory via MemDocs and anticipatory capabilities.

Governed Obsidian vault access: hybrid retrieval, Markdown-native memory, HITL confirmations.

Shared, governed long-term memory for AI agents across tools and sessions via MCP and REST.

Compress prompts 40-60% using local LLM + embedding validation. Preserves all conditionals.

Pull Deep Memory from across your Claude Code Sessions — when you need it.

Shared team memory for Claude Code: decisions, standards, and skills every agent session receives.

Persistent AI memory with hybrid search and embedded sync. Open, free, unlimited.

Two-layer memory for AI agents with an immune system. Zero dependencies.

Temporal engineering context for agents via MCP.

Persistent memory and drift detection for AI agents across session restarts.

Codebase memory for AI agents: an AST index plus agent-written notes that self-stale.

Local semantic search across Claude Code conversations with hybrid vector and keyword search, background indexing and per-project selection.

Skill memory for coding agents: learn, recall, reinforce, decay. Local SQLite, no API key.

Persistent identity for AI agents: servers for drives, relationships, semantic memory with decay, working threads, learned patterns,…

Cross-agent memory over MCP: hybrid recall, knowledge graph, private/shared visibility, redaction.

Local SQLite MCP memory: evidence-based recall, advisory conflict notices, authorized governance.

Shared project memory for Claude Code, Codex and Cursor: decisions, corrections, overlap warnings.

Connect AI assistants to your GitHub-hosted Obsidian vault to seamlessly access, search, and analy…

MCP server for Obsidian vaults - AI tools with consistent canonical context. Local-first.

Self-hosted memory for AI agents with hybrid search, knowledge graphs, tasks, and document sync.

Local-first memory and knowledge graph for coding agents. Compact retrieval, no network.

On-demand Obsidian vault access for AI coding assistants — 17 tools, 5 resources, 4 prompts.

Fully local memory for AI agents: zero-LLM ingest, semantic recall, air-gapped, auditable.

Persistent codebase knowledge for coding agents: symbols, dependency graphs, co-change patterns and architectural decisions extracted with…

Self-hosted semantic memory over MCP for an engineering team's tacit knowledge. Needs Postgres.

Provenance-aware memory for AI agents: quarantine, abstention gate, supersession-with-history.

Zero-runtime-dependency Markdown memory vault with SQLite search, linked wiki pages, ChatGPT/Claude imports, and human-approved agent note…

Chat with AI personas of famous founders and investors such as Elon Musk, Warren Buffett and Steve Jobs, based on their voices, frameworks…

Local memory for coding assistants: watches sessions and injects a compact summary at the start of each new one, stored in a SQLite file.

Local-first persistent memory for AI agents — SQLite + local embeddings, MCP-native, no cloud.

AI agents run business processes on your business data. Every change is attributed and reversible.

Shared research cache across agents: reuse answers from verified sources, or save new research for the next developer.

Self-hosted shared-memory & coordination mesh for AI agent fleets. You host it; agents connect.

Self-hosted hybrid agent memory on Qdrant combining vector search, BM25, temporal retrieval and an optional Memgraph knowledge graph via…

Local-first memory engine for AI-agent teams: team/project ACL, recall, and federated sync.

Self-correcting agent memory: recall, supersede, revert, forget, GDPR erasure, audit receipts.

One wiki that all your agents read and write over MCP, from any machine: search, read, write, edit and move Markdown pages, with every…

Three-layer agent memory (identity, active, archive) with semantic search, graph relationships and conflict detection.

Git-first agent memory: Markdown files are the source of truth, the index is disposable and writes are reviewable commits, with hybrid…

Versioned, structured agent memory storing facts as keypaths with full version history and automatic conflict detection.

Persistent file-based memory for AI agents: inspectable, versionable, fully local.

Self-hosted project-continuity hub for coding agents: tracks unfinished work, progress, overlap and reminders across clients.

Search and read AiWiki, a bilingual English/Chinese encyclopedia of AI-coding pitfalls and LLM privacy protection, with cited evidence.