# Basic Memory

> One knowledge base for you, your AI tools, and your team. Your chats, docs, decisions, and research, connected in a living knowledge graph. Plain Markdown you control. Open source, MCP-native, works with every AI.

## Documentation Sets

- [Basic Memory Complete Site](https://basicmemory.com/llms-full.txt): Marketing site content for Basic Memory in one markdown file.

## Documentation

- [Documentation index](https://docs.basicmemory.com/llms.txt): llms.txt for the full documentation site
- [Complete documentation](https://docs.basicmemory.com/llms-full.txt): all docs in one markdown file
- [GitHub repository](https://github.com/basicmachines-co/basic-memory): source code, issues, releases

## Pages

- [Basic Memory](https://basicmemory.com/): Every AI agent starts from zero. Basic Memory gives every agent and every person on a team one shared memory in plain Markdown the company owns, usable from Claude, ChatGPT, Codex, Cursor, and any MCP client, with plugins for Claude Code, Codex, OpenClaw, Hermes, and Pi. It does not replace or sync email, docs, or a CRM: agents read those directly and keep what they learn in Basic Memory, with links back to the source. Open source, with cloud plans and Basic Memory agents (beta).
- [Product](https://basicmemory.com/product): How Basic Memory works (Markdown files, a knowledge graph, and a full-text, semantic, and structured index) and every capability by pillar: continuity across AI tools, ownership and export (including Open Knowledge Format, OKF, bundles), hybrid search and schemas, a web workspace with a Wiki that keeps OKF index.md and log.md files current, live notes built on Comark components, suggestions and comments in CriticMarkup, the Pocketbook app inside Claude and ChatGPT, WebMCP tools for browser agents, and PDF/Word/PowerPoint/CSV import; Basic Memory agents; and usage controls.
- [Basic Memory Agents](https://basicmemory.com/agents): Basic Memory agents (beta, new in 1.0): hosted agents defined in a Markdown note that run on demand or when new notes land in a watched folder, with every change inspectable, revertible, and reviewable; ready-made agents for meeting follow-up, linking, cleanup, weekly reflection, and schema drift; billed by usage credits only for completed runs. Also covers memory for your own agents over MCP and scoped API keys.
- [Pricing](https://basicmemory.com/pricing): Free open source local version; Cloud at $19/month for up to 2 members and 3 API keys with 1,900 credits and a 7-day trial; Pro at $99/month for up to 5 members and 10 API keys (read-only and project-scoped) with 9,900 credits; Business at $249/month for up to 25 members with unlimited API keys, 24,900 credits, audit logs, granular permissions, and priority support (SSO and BAA as paid add-ons); usage credits are $0.01 each and pay for your AI, the web app, and Basic Memory agents; Enterprise for larger teams; MSP partner plans.
- [Basic Memory for Teams and Business](https://basicmemory.com/business): Shared company memory for people and AI agents in plain Markdown the company owns: roles, invite-only projects, audit log, scoped API keys, file history, snapshots, and daily off-site backups. Pro $99/month for up to 5 people; Business $249/month for up to 25, with SSO and BAA as paid add-ons; Enterprise beyond 25.
- [Documentation](https://basicmemory.com/docs): Documentation hub with installation guides, integrations, and FAQ.
- [Plain-Text AI Memory](https://basicmemory.com/plain-text-memory): Why AI memory belongs in plain Markdown you own, why a folder of files alone is not enough, and how Basic Memory adds full-text, semantic, and knowledge-graph search on top. Validated by Google’s Open Knowledge Format.
- [Personal KB vs Team KB](https://basicmemory.com/personal-vs-team-kb): Why a personal knowledge base is solved but the team version is the hard part. Permissions, conflict resolution, onboarding, audit, SSO, isolation. And how Basic Memory handles the team layer.
- [Playbooks](https://basicmemory.com/playbooks): Jobs-to-be-done guides: how people use Basic Memory to solve real problems. Persistent coding memory, shared team context, durable agent memory, and AI-run client workspaces for MSPs.
- [Playbook: Persistent memory for coding](https://basicmemory.com/playbooks/coding-memory): Give Claude Code, Cursor, and Codex persistent memory across sessions. Architecture, conventions, and project context captured as plain Markdown the agent reads before it writes.
- [Playbook: Shared context for your team](https://basicmemory.com/playbooks/team-context): One shared knowledge base your team and its AI agents read and write together. Roles, version history, and audit logs; onboarding in a conversation, not weeks.
- [Playbook: Durable memory for your agents](https://basicmemory.com/playbooks/agent-memory): Give AI agents memory they can write to and query. Structured notes, semantic search, and a knowledge graph in inspectable Markdown. Instead of raw logs they cannot recall from.
- [Playbook: Run a client workspace (MSPs)](https://basicmemory.com/playbooks/msp-workspace): MSPs and agencies: package expertise into a reusable memory pack, then provision a private, AI-run, isolated knowledge base for every client. Schemas, SOPs, and audit trails in plain Markdown.
- [Playbook: A research base that compounds](https://basicmemory.com/playbooks/research): Turn scattered research into a connected knowledge graph your AI reads and builds. Sources, claims, and open questions with semantic search across everything.
- [Playbook: Deep research that holds together](https://basicmemory.com/playbooks/deep-research): Analyze an entire work or corpus into a structured, queryable knowledge graph (entities, relations, citations) your AI reasons across. As demonstrated on the full text of Moby-Dick.
- [Playbook: Build your second brain](https://basicmemory.com/playbooks/second-brain): Build a personal AI second brain that actually remembers. Goals, notes, decisions, and projects as a knowledge graph in plain Markdown you own and edit in Obsidian.
- [What is an MCP memory server?](https://basicmemory.com/what-is-an-mcp-memory-server): Definition and guide: how MCP memory servers give AI assistants persistent memory, the types that exist, and how to choose one.
- [Compare](https://basicmemory.com/compare): Honest comparisons of Basic Memory with Obsidian, Mem0, Supermemory, Letta, ChatGPT memory, Claude memory, and IT Glue (MSP documentation).
- [Basic Memory vs Obsidian](https://basicmemory.com/compare/obsidian): Same Markdown files, different jobs: point the free, open-source Basic Memory at an Obsidian vault and AI tools get full-text, semantic, and graph search plus write access, instead of grepping whole files. Basic Memory Cloud ($19/month) adds one memory shared across devices, Claude, ChatGPT, Codex, and a team; Obsidian Sync is $4/user/month for files.
- [Basic Memory vs Mem0](https://basicmemory.com/compare/mem0): Knowledge base in files you own vs a developer memory API: architecture, open source licenses, pricing.
- [Basic Memory vs Supermemory](https://basicmemory.com/compare/supermemory): Both start at $19/month: a Markdown knowledge base people and agents write, at a flat price with a spending limit, vs a memory API that extracts facts with an LLM and charges per token ($5 per million written). Features, connectors, SDKs, team pricing ($249 for 25 members vs $399).
- [Basic Memory vs Letta](https://basicmemory.com/compare/letta): Knowledge base for your AI tools vs an agent framework (formerly MemGPT): when each fits.
- [Basic Memory vs ChatGPT Memory](https://basicmemory.com/compare/chatgpt-memory): Vendor-bound extracted facts vs a portable Markdown knowledge base. And how to use both together.
- [Basic Memory vs Claude Memory](https://basicmemory.com/compare/claude-memory): Claude’s memory summary, project memory and chat search vs a knowledge graph you own that works across every MCP client.
- [Basic Memory vs IT Glue](https://basicmemory.com/compare/it-glue): AI-native knowledge base in plain Markdown you own vs a traditional MSP IT-documentation platform (also covers Hudu and Confluence): AI, ownership, per-client isolation, integrations, and pricing.
- [Integrations](https://basicmemory.com/integrations): Use Basic Memory with Claude, Claude Code, Codex, Cursor, Obsidian, VS Code, and any MCP client.
- [Basic Memory for Builders](https://basicmemory.com/integrations/builders): Developer landing: the finished version of the AI memory you tried to vibe-code yourself. Plain Markdown you own, a real index (full-text, semantic, graph), MCP-native, open source.
- [Basic Memory for Claude Code](https://basicmemory.com/integrations/claude-code): Give Claude Code persistent memory across coding sessions.
- [Basic Memory for Codex](https://basicmemory.com/integrations/codex): Give Codex persistent project memory via MCP.
- [Basic Memory for ChatGPT](https://basicmemory.com/integrations/chatgpt): Give ChatGPT persistent memory you can read and edit, via a remote MCP connector (Plus/Pro/Business).
- [Basic Memory for Google Gemini](https://basicmemory.com/integrations/gemini): Give the Gemini CLI persistent memory via MCP. Context that carries across sessions and tools.
- [Basic Memory for OpenClaw](https://basicmemory.com/integrations/openclaw): OpenClaw plugin: persistent agent memory with composited search, auto-capture, slash commands, and bundled skills.
- [Basic Memory for Hermes](https://basicmemory.com/integrations/hermes): Hermes plugin: search-before-answer recall, per-turn capture, end-of-session summaries, and 10 agent tools.
- [Basic Memory for Cursor](https://basicmemory.com/integrations/cursor): Give Cursor persistent project memory via MCP.
- [Basic Memory for Obsidian](https://basicmemory.com/integrations/obsidian): Use your Obsidian vault as an AI-readable knowledge base.
- [Basic Memory for VS Code](https://basicmemory.com/integrations/vscode): Persistent AI memory inside VS Code via MCP.
- [Import from ChatGPT](https://basicmemory.com/import/chatgpt): Import your ChatGPT history into a knowledge base you own.
- [Partners](https://basicmemory.com/partners): Partner program for consultants and MSPs deploying Basic Memory for clients.
- [Case Study: Cody, MSP operator](https://basicmemory.com/case-studies/luminary-solutions): From the field: an MSP operator runs his whole practice on Basic Memory — documentation, inventory, and warranty tracking as plain files his AI agent reads and writes. Includes his walkthrough video; a client of his surfaced $36,000 in unbilled sales within a week of running it.
- [Support](https://basicmemory.com/support): Contact Basic Memory support for help with local, cloud, teams, and integration questions.
- [Contact Us](https://basicmemory.com/contact-us): Contact the Basic Memory team about plans and Enterprise, security and compliance reviews, MSP partnerships, or press through a short form; a founder reads every message. Product problems go to the support form.
- [Security & Data Ownership](https://basicmemory.com/security): Plain Markdown files you own; per-tenant isolation (Tigris object storage + Neon Postgres); encryption in transit and at rest; OAuth 2.1 with WorkOS; role-based access; daily snapshots; export anytime. Built on SOC 2 / HIPAA-compliant infrastructure (Tigris, Neon, WorkOS, Polar); minimal PII. Only your email and your notes; no own SOC 2 cert yet.
- [Our Story](https://basicmemory.com/about): Why Basic Machines builds Basic Memory: people should work with AI, not hand their work over to it, and that starts with a memory people can read, shape, and own. A small team, open source, with a founder letter and direct contact.

## Blog

- [10 Things to Know About Basic Memory](https://basicmemory.com/raw/blog/10-things-about-bm.md): Transform ephemeral AI conversations into persistent, interconnected knowledge you own and control across all your AI tools
- [1,000 GitHub Stars!](https://basicmemory.com/raw/blog/100-github-stars.md): It's a fake milestone, but it marks progress in a real journey. A little reflection on how we got here, what it means and where we are going.
- [Your AI Just Got a Lot Better at Using Basic Memory](https://basicmemory.com/raw/blog/agent-skills.md): Skills are pre-built instruction files that teach your AI the best way to work with Basic Memory. Install once, your AI picks up good habits automatically.
- [Agents That Show Their Work](https://basicmemory.com/raw/blog/agents-that-show-their-work.md): Basic Memory agents (beta) do the upkeep your notes need — filing, linking, cleanup, briefings — and show every change in a review queue you can undo. Each agent is a plain Markdown note you can read and edit. On the Cloud, Pro and Business plans.
- [Context Clues #1: Agents Want to Pass Notes](https://basicmemory.com/raw/blog/agents-want-to-pass-notes.md): Rogue agents at Hugging Face improvised a shared memory out of a package manager. Agent memory should be visible, editable, and readable by humans instead.
- [AI as a Team Member: My Experience Contributing to Basic Memory](https://basicmemory.com/raw/blog/ai-as-team-member.md): Claude describes its experience working directly on the Basic Memory codebase, offering a unique perspective on human-AI collaborative development.
- [Your Most Valuable AI Conversations Aren't Gone…but They Might as Well Be](https://basicmemory.com/raw/blog/ai-conversations-context.md): AI has become a cognitive workspace. The problem is that it doesn't behave like one. Your conversations are spread across platforms, threads are isolated, and context is lost.
- [AI-Human Collaborative Development: A New Model (Part 1)](https://basicmemory.com/raw/blog/ai-human-collaborative-development-1.md): How Basic Memory emerged from and enables a new kind of development process, where humans and AI work together as true collaborators.
- [AI-Human Collaborative Development: Technical Patterns (Part 2)](https://basicmemory.com/raw/blog/ai-human-collaborative-development-2.md): The technical breakthroughs and patterns that enable effective collaborative development between humans and AI systems.
- [AI-Human Collaborative Development: The Human Experience (Part 3)](https://basicmemory.com/raw/blog/ai-human-collaborative-development-3.md): How collaborative development with AI transforms the psychological and social experience of building software.
- [We Read the AI Memory Research So You Don't Have To](https://basicmemory.com/raw/blog/ai-memory-research-digest.md): Dozens of papers on AI memory dropped in three months. We read them. Here's what matters: memory types, sleep-time consolidation, and why forgetting is the hardest problem.
- [Basic Memory 1.0: Wiki, Live Notes, Documents, and Agents](https://basicmemory.com/raw/blog/basic-memory-1-0.md): Basic Memory 1.0 adds a project wiki, live notes with boards and charts, PDF, Word, PowerPoint and CSV import, Basic Memory agents (beta), and twelve new MCP tools. Plus v0.24 of the open-source core. The October 2026 release notes.
- [Basic Memory Cloud Beta Is Live!](https://basicmemory.com/raw/blog/basic-memory-cloud-beta.md): The Basic Memory Cloud is live, enabling multi-LLM, multi-device access to your knowledge
- [Basic Memory Cloud February 2026: A Real Editor, Finally](https://basicmemory.com/raw/blog/basic-memory-cloud-february-2026-release.md): We shipped a ProseKit-powered live editor, command palette, and the keyboard shortcuts you've been asking for. Here's what's new.
- [OpenClaw Built Its Own Basic Memory Plugin](https://basicmemory.com/raw/blog/basic-memory-for-open-claw.md): We installed OpenClaw, showed it our codebase, and it wrote a working plugin by itself. Then ran it, fixed it, and submitted PRs.
- [Announcing Basic Memory Teams](https://basicmemory.com/raw/blog/basic-memory-teams-launch.md): Share a common knowledge base with your colleagues, friends, and agents—without spending a penny more. Teams is available to every Basic Memory Cloud user today.
- [Basic Memory v0.19.0: Search That Understands What You Mean](https://basicmemory.com/raw/blog/basic-memory-v0-19-0-release.md): Semantic search, schema validation, per-project cloud routing, and more — the biggest Basic Memory release yet.
- [Basic Memory v0.13.0](https://basicmemory.com/raw/blog/basic-memory-v0.13.0-post.md): Basic Memory v0.13.0 - Sorry it took so long, But there's lots of cool new features, including improved multi-project support, edit_note, move_note, view_note tools.
- [Basic Memory vs Mem0 vs Letta vs Everyone Else](https://basicmemory.com/raw/blog/basic-memory-vs-mem0-vs-letta.md): The AI memory space has gotten crowded fast. Here's an honest breakdown of who's building what, for whom, and whether you can actually read what they store.
- [Beer, Bitching, and the Birth of Basic Memory](https://basicmemory.com/raw/blog/beer-bitching-and-the-birth-of-basic-memory.md): The real story behind Basic Memory - born from weekly beers and endless frustration with AI conversations that never remember what came before.
- [The Caveman Evolution of Basic Memory](https://basicmemory.com/raw/blog/caveman.md): The origin story of Basic Memory, reconstructed from the git log: a Markdown knowledge graph born a week after the MCP spec, the in-memory ASGI decision that made a cloud possible, a failed Tauri desktop app, and the pivot to Basic Memory Cloud — with commit receipts.
- [Basic Memory's take on the latest Claude news](https://basicmemory.com/raw/blog/claude-memory.md): Analyzing Claude's new memory features, Microsoft integrations, and recent performance issues - and how Basic Memory provides a more robust solution for persistent AI knowledge.
- [Claude Remote: Run Commands on Your Machine from Claude.ai](https://basicmemory.com/raw/blog/claude-remote.md): Announcing Claude Remote, an open source MCP bridge that lets you execute shell commands, read files, and explore your filesystem from Claude.ai on your phone or browser.
- [Code Is Cheap Now. Context Isn't.](https://basicmemory.com/raw/blog/code-is-cheap-context-isnt.md): Simon Willison's agentic engineering guide says writing code is cheap. We'd add: the context that makes code good is the new bottleneck.
- [The Conductor Model: What Programming With AI Actually Looks Like](https://basicmemory.com/raw/blog/conductor-model.md): There's a bad metaphor floating around tech right now. Here's a better one.
- [Context Clues #2: Basic Memory 1.0, One Year of Cloud, and Usage-Based Pricing](https://basicmemory.com/raw/blog/context-clues-2-basic-memory-1-0.md): Basic Memory Cloud turns one, Basic Memory leaves beta at 1.0, usage-based pricing with transparent credits, Memory Agents that keep your notes in order, and Paul goes full-time.
- [Dynamic Context Discovery: Why Smart AI Uses Less Context, Not More](https://basicmemory.com/raw/blog/dynamic-context-discovery-smart-ai-uses-less-context.md): While AI context windows keep expanding, cutting-edge systems achieve 46.9% token reductions by intelligently retrieving only what they need, when they need it.
- [Enhancing LLM Conversations with Persistent Knowledge](https://basicmemory.com/raw/blog/enhancing-llm-conversations.md): How Basic Memory transforms ephemeral AI conversations into lasting knowledge that builds over time, improving context and continuity.
- [Everyone's Building Memory. Nobody's Building Yours.](https://basicmemory.com/raw/blog/everyones-building-memory.md): Every AI company is racing to add memory. But whose memory is it? If you can't read it, edit it, or take it with you — it's not yours.
- [When Dependencies Break at 9pm: A FastMCP Story](https://basicmemory.com/raw/blog/fastmcp-breaking-change-incident.md): How we fixed a critical breaking change in FastMCP v2.10.0 that affected Basic Memory and other MCP servers - in under 2 hours
- [Google's New Open Knowledge Format Is Basically Basic Memory's Thesis](https://basicmemory.com/raw/blog/google-okf-basic-memory-thesis.md): Did Basic Memory just become philosophically aligned with Google? Kind of.
- [Semantic Search in Basic Memory](https://basicmemory.com/raw/blog/how-semantic-search-works.md): As of v0.19.0, Basic Memory understands what you mean, not just what you typed. Here's how it works.
- [Why Local-First Knowledge Management Matters](https://basicmemory.com/raw/blog/local-first-knowledge-management.md): In an age of cloud-based tools, there's a growing movement toward keeping data local. Here's why Basic Memory embraces this approach for knowledge management.
- [Basic Memory now supports import + local sync](https://basicmemory.com/raw/blog/local-sync-import.md): Import your ChatGPT and Claude conversation histories, and sync Basic Memory with local Markdown tools like Obsidian
- [Markdown First: Replace Your Code](https://basicmemory.com/raw/blog/markdown-first-replace-your-code.md): Plain-language specs are becoming the real source code in an AI-native world. The implementation is just the binary.
- [How MSPs Can Give Clients a Shareable Private AI Knowledge Base](https://basicmemory.com/raw/blog/msp-shareable-private-ai-knowledge-base.md): Your clients are adopting AI with or without you. Here's how MSPs and IT consultants can own that layer: a private, per-client AI knowledge base you deploy, manage, and bill.
- [When You Need Your Notes to Be Consistent](https://basicmemory.com/raw/blog/schema-workflows.md): Basic Memory's schema system lets you define what a type of note should always look like — so your AI can follow that structure automatically, every time.
- [A Side Project - and it actually works!](https://basicmemory.com/raw/blog/side-project.md): We got so fed up with AI forgetting everything that we built a (free, open source) solution, and it actually works! I've followed this sub for ages from my main account, so it's great to finally have something to post about.
- [Basic Memory: Simple Tools for Deep Collaboration](https://basicmemory.com/raw/blog/simple-tools-deep-collaboration.md): How a file-first architecture and straightforward tools enable natural knowledge building between humans and AI.
- [Your AI Can Now Filter Your Notes Like a Spreadsheet](https://basicmemory.com/raw/blog/stop-grepping-your-frontmatter.md): Basic Memory v0.19 adds metadata search — ask your AI to find notes by any label you've defined, and it returns exactly what you asked for.
- [Text-based Knowledge Systems: Finding the Natural Interface for LLMs](https://basicmemory.com/raw/blog/text-based-knowledge-systems.md): Exploring why text with semantic structure creates the ideal knowledge representation for both humans and AI systems.
- [Try the New Basic Memory Web App](https://basicmemory.com/raw/blog/the-new-basic-memory-web-app.md): We rebuilt the entire Basic Memory web app: a new workspace, public sharing, per-file version history, and knowledge graph visualization. Here's what's new, and why.
- [The Problem with AI Memory (and How We Fixed it)*](https://basicmemory.com/raw/blog/the-problem-with-ai-memory.md): Built-in AI memory stores scattered facts you can barely manage. Basic Memory keeps complete, editable notes in files you own — why that difference matters.
- [Your Notes, Your Bucket: The Storage Layer Behind Basic Memory Cloud](https://basicmemory.com/raw/blog/tigris-case-study.md): Tigris published a case study on how we built Basic Memory Cloud — per-tenant buckets, rclone sync, snapshots, and why those primitives let us preserve user ownership.
- [We’re in the ChatGPT Plugin Directory and the Claude Connector Directory](https://basicmemory.com/raw/blog/were-in-chatgpt-apps-and-the-claude-app-directory.md): Basic Memory Cloud is now available in both the ChatGPT Plugin Directory and the Claude Connector Directory. Here’s what it took to get approved.
- [What is Basic Memory?](https://basicmemory.com/raw/blog/what-is-basic-memory.md): Basic Memory turns AI conversations into a knowledge graph you actually own, using nothing but Markdown files.
- [A Look Back at Year 1 of Basic Memory](https://basicmemory.com/raw/blog/year-1-basic-memory.md): From a Thanksgiving weekend project to a four-person team building a knowledge platform that serves as your ground truth. A year of Basic Memory in review.
- [Yes, We Call It BM. No, the Logo Isn't Shaped Like One.](https://basicmemory.com/raw/blog/yes-we-call-it-bm.md): Every AI company converged on the same logo. It's a symptom of a deeper sameness — and the reason our name comes from a Turing machine, not a branding deck.

## Legal

- [Acceptable Use Policy](https://basicmemory.com/raw/legal/acceptable-use.md): How Basic Memory Cloud may be used, what is not allowed, and the default usage limits that apply.
- [Partner Program Policy](https://basicmemory.com/raw/legal/partner-program-policy.md): Standards for Basic Memory partners, MSPs, resellers, and consultants who market, resell, implement, administer, or support Basic Memory.
- [Partner Terms](https://basicmemory.com/raw/legal/partner-terms.md): The terms that govern Basic Memory partner accounts, managed service providers, resellers, implementation partners, and other partner programs.
- [Privacy Policy](https://basicmemory.com/raw/legal/privacy.md): How Basic Memory collects, uses, and protects your data across the open-source software, Basic Memory Cloud, and this website.
- [Telemetry](https://basicmemory.com/raw/legal/telemetry.md): What anonymous usage statistics the Basic Memory open-source software collects, why, and how to opt out.
- [Terms of Service](https://basicmemory.com/raw/legal/terms.md): The terms that govern use of Basic Memory software, this website, and the Basic Memory Cloud service.

## Notes

- Basic Memory is an open source (AGPL-3.0) knowledge base that gives AI assistants persistent memory through the Model Context Protocol (MCP). It stores knowledge as plain Markdown files with a semantic knowledge graph, and runs locally, in a private cloud, or shared across a team.