Give MCP clients local Markdown project memory with Basic Memory
Connect Claude, Codex, Cursor, ChatGPT, or any MCP-capable client to a local-first Markdown knowledge graph so agents can write, search, and reuse project context across sessions.
npx skills add agentskillexchange/skills --skill give-mcp-clients-local-markdown-project-memory-with-basic-memory
Use Basic Memory when an operator wants an AI coding or research session to retain durable project knowledge without replaying entire chats or relying on a hosted product as the source of truth. The agent workflow is concrete: install the Basic Memory CLI, connect it to an MCP-capable client, let the agent write observations and linked notes into local Markdown, search those notes semantically in later sessions, and curate the resulting project memory files. Invoke it when the missing job is persistent, inspectable project memory for agent work across sessions. The boundary is the MCP/local-Markdown memory workflow, not general note-taking, a generic cloud knowledge base, or a broad memory SDK.
What this skill actually does
Inputs and prerequisites: Python 3.12+, uv, Basic Memory CLI, an MCP-capable client such as Claude, Codex, Cursor, ChatGPT, or VS Code, and a local Markdown workspace.
Setup notes: Install with `uv tool install basic-memory`, create or select a local Basic Memory project, then configure the Basic Memory MCP server in the target AI client so the agent can write, search, and fetch Markdown-backed memories.
Source and verification boundary: use https://github.com/basicmachines-co/basic-memory#readme as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a MCP workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.