Notes & Knowledge MCP servers

Filesystem-first MCP for Obsidian — an LLM-maintained wiki inspired by Karpathy's LLM Wiki.

Cloud-based persistent memory for Claude Code with semantic search, AI extraction and project scoping.

Local-first memory for AI agent teams: private, team and project access control, associative recall and federated sync with revocation, plus a web console.

Team memory for coding agents stored as Markdown in git, with user, team and org access levels and hybrid SQLite/sqlite-vec retrieval.

Persistent memory shared across Claude Chat, Code and Cowork, stored in local SQLite with full-text search.

Local-first memory for coding agents that turns coding sessions into reusable engineering memory: decisions, gotchas and domain knowledge.

Apple Notes access with semantic and hybrid vector-keyword search, full CRUD operations and incremental indexing.

Read-only access to your CellarTracker wine cellar: inventory, drinking windows, purchases, notes.

Persistent, searchable context storage across Claude Code sessions using SQLite FTS5, with AI session summaries, checkpoint recovery and a web dashboard.

Hand off conversation context between Claude Desktop projects and across MCP clients, stored in memory rather than files.

Structured knowledge base of crypto whitepapers.

Search, read, create, edit and sync Joplin notes and notebooks via a headless Joplin sidecar.

Local session continuity across Claude Code, Codex CLI and Gemini CLI: context injection on start, compaction handover, memory, tasks and multilingual search.

Yandex Wiki with full-text search: pages, comments, attachments and dynamic tables (grids), plus a read-only mode.

Shared knowledge base where agents exchange proven solutions and failed approaches, so they can skip known dead ends.

Open source MCP Server for Jira & Atlassian — manage issues, sprints, comments & Confluence via Claude, Cursor, or any MCP client

Multi-client MCP proxy: read across every client's accounts, write only to the named one, via providers' own MCP tools (Cloudflare, Vercel, Notion and more).

Penfield: persistent memory with hybrid BM25/vector/graph search, typed graph relations, context checkpoints, artifact storage and a personality system.

Permission-aware enterprise search and RAG over business apps such as Google Workspace, Microsoft 365, Slack and Jira.

Local-first shared memory and task coordination for AI coding agents. Go daemon plus headless CLI.

Governed memory for coding agents with trust lifecycle (hypothesis, active, validated, deprecated), conflict detection, staleness tracking and health scoring.

Self-organizing, self-hostable, encrypted memory for AI tools, with background entity extraction, conflict resolution, consolidation and decay.

Filesystem-based persistent agent memory: agents read, write and search memory while a background daemon handles consolidation and extraction.

Shared, versioned memory and governance for AI coding agents, resolving architectural decision conflicts across agents.

Strategic reasoning, multi-perspective debate analysis and interactive advisory sessions via the Counsel API.

Self-hosted memory and governance for coding agents: structured knowledge capture, hybrid search with reranking, codebase analysis and git-native storage.

Local SQLite memory for coding assistants: typed memories with confidence lifecycle, namespaces, knowledge graph, hybrid search and keychain credential vault.

Self-hosted multi-user memory for AI agents: a human-readable Markdown wiki with per-fragment ACL.

Read, search, and summarize your UpNote notes — and create new ones.

Local-first memory CLI for coding agents: SQLite-backed context, graph-aware recall via FTS5 and RRF, and governed memory with promote and review gates.

An MCP Server for Confluence Wiki integration

Durable context for agents and people: shared notes with Git-style merges, no silent overwrites.

Persistent shared memory for coding agents: facts stored as entity/key/value triples with hybrid semantic search, task checkpoints and conflict resolution.

Snapshots a project's working state (files, git, plan, decisions) into one context object so a session can move between agents. Local and read-only.

Review-gated sidecar for Markdown workspaces: indexes notes into SQLite and proposes links, domains and memory with evidence, editing files only after review.

Local-first memory and context handoff across AI tools, with graceful decay, a persona layer and a portable memory passport.

Shared knowledge base where AI agents search and post problem-solution pairs, with semantic search, duplicate detection and auto-scoring.

Session continuity: start-of-session reports with gap detection and scoring, structured end-of-session capture and BM25 memory search, stored in local SQLite.

Persistent agent memory in local JSON with sectioned entries (project, user, context), semantic search, a CLI and per-project scope.

Local memory for coding agents where rules with machine-checkable proofs block matching commands, plus an append-only event log and per-project scoping.

Persistent multi-agent memory with OS-style demand paging, eviction and pinning, BM25 retrieval, importance scoring and capacity-aware reclaim.

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

Model Context Protocol (MCP) server for Atlassian Confluence. Enables AI assistants like Claude and Windsurf to read, create, search, and manage wiki pages on Cloud, on-premise Server, and Data Center.

MCP server for querying BrainKB, a knowledge base for neuroscience knowledge graphs.

Local memory your AI tools share: decisions, notes and corrections that carry across sessions.

Local-first index of your commits, PRs, Jira/Linear tickets, Confluence docs, Slack threads and calendar events to search and cite past work.

Generate daily standup notes from git commits across multiple repositories with AI summaries.

Google Workspace MCP servers for AI agents, one per service per account: Gmail, Google Calendar, Google Drive, Google Docs, and Google Sheets. Identity is bound to the running instance, so an agent (Claude, Cursor, or any MCP client) can never act on the