MemoryAtlas

Local-first knowledge-graph memory layer for AI agents and humans, exposed entirely via MCP. Conversations and notes are stored as plain Markdown files; observations and wikilinks compound into a semantic graph over time. Designed to work with any AI client or IDE that speaks MCP — Claude, Copilot, Cursor, and others. A team cloud tier (basicmemory.com) provides shared workspaces.

Storage
Markdown files on local disk are the source of truth — every note, observation, and entity is a human-readable file. Wikilinks between notes build a knowledge graph that accumulates automatically as agents write memories. All writes go through MCP tools so no direct file-system access is required from the agent.
Retrieval
Semantic search and graph traversal via MCP tools. Agents query the knowledge graph through a standardised MCP server; the server resolves wikilinks to expand context and uses embedded similarity for fuzzy recall. The local setup requires no external embedding API (uses a bundled model).
Self-host
Self-host: trivial
License
AGPL-3.0
Pricing
Open-source AGPL-3.0, free to self-host (`uv tool install basic-memory`). Cloud tiers at basicmemory.com/pricing: Team $15/seat/mo (volume discounts to $12/seat at 20+ seats), Business $30/seat/mo (adds SSO/SAML, audit logs, HIPAA-compliant infra), and custom credit-metered Agent Infrastructure for fleets/on-prem; 7-day free trial on paid tiers. · Free + paid
GitHub stars
3,616
Last release
2026-06-13
Last commit
2026-08-10
First catalogued
2026-06-28

Strengths

  • Zero-infra local-first install (single `uv` command); runs entirely on local disk with no external API keys required
  • MCP-native: works with any agent or IDE that supports the Model Context Protocol out of the box
  • Human-readable Markdown + wikilink graph — memories are inspectable and editable by both humans and AI
  • Active development with frequent releases; 3330+ stars; AGPL-3.0 open-source core

Watch out

  • AGPL-3.0 license may require open-sourcing application code in commercial products that incorporate it
  • Cloud tiers (Team $15/seat/mo, Business $30/seat/mo) are priced per seat and billed monthly by default (a yearly/discounted rate applies at higher seat counts) — model per-seat cost before assuming it's cheaper than self-hosting
  • Knowledge graph quality depends on consistent MCP-tool usage; agents that bypass MCP and write files directly will miss graph updates
  • Default install emits anonymous telemetry (cloud promo impressions) to an external Umami Cloud instance — set BASIC_MEMORY_NO_PROMOS=true for true air-gapped environments

Best for

  • Individual developers and small teams wanting persistent cross-session memory for AI coding assistants with zero infrastructure
  • Projects where human-readable memory files and direct editing are a design requirement

How it integrates

Benchmark results

No sourced results yet.

Sources

Last verified 2026-08-01 · updated by refresh-framework-cards