Basics
Enables importing community packs from GitHub repositories, with new notes and updates arriving as inbox proposals for review and approval.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Basicsrecommend a stack for a REST API serving ML models"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Basics
English | Español
Basics is a personal tech-stack vault: a folder of Markdown notes about the layers, technologies, practices, checks and
profiles you use, which your coding agents read and extend over MCP. Notes are plain files (frontmatter +
[[wikilinks]], Obsidian- and git-friendly), and every agent write lands in an inbox for you to approve.
Status: 1.0 — see the stability contract.
Quick start
# create a vault with the official packs
npx @marcozvd/basics init ~/basics-vault --yes
# open the local web UI (it also serves MCP over HTTP)
npx @marcozvd/basics --vault ~/basics-vault uiWith the UI running, connect Claude Code:
claude mcp add --transport http basics http://127.0.0.1:4747/mcpAsk your agent to recommend a stack, then review and approve what it proposes in the Inbox. Other clients (Cursor, VS Code, Codex, OpenCode, Windsurf) are covered in MCP clients.
Related MCP server: memory-vault
What it does
A vault you can read — layers, tech, practices, checks, profiles and projects as Markdown notes. No database, no lock-in.
Agents that ask first — agent writes become proposals in
inbox/with a reason and a diff; you approve in the CLI or the UI.Recommended stacks — rank technologies per layer from a project description, by keywords or with optional local embeddings.
From stack to project — export
STACK.md,AGENTS.md,CLAUDE.mdand Cursor rules, then run the stack checks in the project.Packs — official starter packs in English and Spanish, plus community packs from any GitHub repo; new notes and updates arrive as inbox proposals, never as silent writes.
Local web UI and MCP — a React UI with editor, graph, inbox diff and recommendation playground, plus MCP over HTTP or stdio.
Desktop app — the same UI and MCP server running in a standalone, tray-resident window with its own bundled Node runtime; no npm needed. See Desktop app.
Agent skill — a
SKILL.mdthat teaches Claude Code, OpenCode or Codex when to call each Basics tool. See Agent skill.
Screenshots (web UI, Inbox diff, graph): coming soon.
Documentation
Full documentation: marcozvd.github.io/basics
Start — Getting started · Concepts
Use — Notes · Packs · MCP clients · Web UI · Recommend · Projects and checks
Operate — Security · Troubleshooting · Stability
Reference — CLI · MCP tools · Vault schema
Contributions: CONTRIBUTING.md. Security reports: SECURITY.md.
License
MIT — see LICENSE.
This server cannot be deployed
Maintenance
Related MCP Connectors
Agent-native notes, tasks, dev-docs, vaults, sync & handoffs. MCP + OpenAPI dual surface.
- WitWikiOAuthapp.witwiki
A shared team wiki your coding agents read and write — across every repo and every MCP client.
Shared long-term memory vault for AI agents with 20 MCP tools.
Team docs served to AI agents over MCP - search, Markdown reads, version pinning, read audit.
Related MCP Servers
- AlicenseNot gradedqualityAmaintenanceProvides persistent knowledge capture and retrieval for coding agents. Enables searching the vault, storing notes, capturing sessions, and reading notes via MCP tools.13MIT
- AlicenseAqualityBmaintenanceProvides cross-session persistent memory for coding agents via MCP tools to store, retrieve, and manage notes with hybrid keyword/semantic search and automatic deduplication.8201 npmMIT
- FlicenseAqualityCmaintenanceProvides coding agents with a shared, Markdown-based engineering knowledge base to search, capture, create, and update internal conventions, API details, infrastructure configs, and development setup via lightweight MCP tools.6-
- FlicenseNot gradedqualityAmaintenanceEnables AI assistants to save and retrieve code snippets, perform semantic vector search, generate code style guides and wiki documentation, and orchestrate multi-agent workflows through MCP tools.-