Anthropic Auto Dream vs Official MCP Memory server

Anthropic Auto Dream vs Official MCP Memory server: side-by-side comparison of two claude code memory mechanisms systems — architecture, taxonomy, license, pricing, MCP/A2A support, and direct edges.

Anthropic Auto Dream · Official MCP Memory server

Cost & capability

Anthropic Auto DreamOfficial MCP Memory server
Capability bandcompetententry
Capability composite7035
Cost tiermidfree
$/Mtok input0
$/Mtok output0
Use casesCode Generation Focused, Long Running SessionCode Generation Focused, Long Running Session, Offline Capable, Multi Agent Coordination

Where they differ (14)

Rows where both sides have data and the values disagree — the shortlist of dimensions that actually distinguish these two systems.

Anthropic Auto DreamOfficial MCP Memory server
Capability bandcompetententry
Capability composite7035
Cost tiermidfree
Use casesCode Generation Focused, Long Running SessionCode Generation Focused, Long Running Session, Offline Capable, Multi Agent Coordination
TypeREM-style sub-agent memory consolidationKG-based persistent memory MCP
Created2026-05 (research-preview rollout in Claude Code; server-side feature flag)2024-11
Pricingbundled with Claude Code (no separate fee)free / open source (MIT)
Backend storagelocal filesystem (~/.claude/projects/<project>/memory/)file (JSON)
DeploymentLocal CLI (Claude Code) — Claude.ai web not in scopelocal (Node.js MCP server; local JSON file)
API surface/dream slash command + /memory toggle (Auto-dream: on)MCP (stdio); knowledge-graph tools (create_entities/relations/observations, read/search/delete)
MCPnative (first-party — Claude Code built on MCP)native (first-party) — reference implementation
OpenTelemetryno — local CLI featureno
Optimised forbackground memory consolidation + bounded MEMORY.md size + surgical updatesClaude-Code-native + MCP-portable memory
Anti-fitnot for users on Claude.ai web (CLI-only); not for environments where consolidation logic must be auditablenot for non-Claude-Code clients (some are MCP-portable)

At a glance

Anthropic Auto DreamOfficial MCP Memory server
SectionClaude Code memory mechanisms Claude Code memory mechanisms
TierT1 T1
TypeREM-style sub-agent memory consolidation KG-based persistent memory MCP
Created2026-05 (research-preview rollout in Claude Code; server-side feature flag) 2024-11
Latest release 2026.1.26 2026-01-27
License Custom
GitHub 85.1k★ +4.8k/mo TypeScript
Pricingbundled with Claude Code (no separate fee) free / open source (MIT)
FundingAnthropic total $12.4B+ raised; $40B valuation
Backend storagelocal filesystem (~/.claude/projects/<project>/memory/) file (JSON)
DeploymentLocal CLI (Claude Code) — Claude.ai web not in scope local (Node.js MCP server; local JSON file)
API surface/dream slash command + /memory toggle (Auto-dream: on) MCP (stdio); knowledge-graph tools (create_entities/relations/observations, read/search/delete)
MCPnative (first-party — Claude Code built on MCP) native (first-party) — reference implementation
A2Anot supported not supported
OpenTelemetryno — local CLI feature no
Optimised forbackground memory consolidation + bounded MEMORY.md size + surgical updates Claude-Code-native + MCP-portable memory
Anti-fitnot for users on Claude.ai web (CLI-only); not for environments where consolidation logic must be auditable not for non-Claude-Code clients (some are MCP-portable)

Taxonomy

AxisAnthropic Auto DreamOfficial MCP Memory server
storagefilegraph
retrievalextraction-pullgraph-traversal
persistencelong-termcross-session
updateconsolidationoverwrite
unitfactfact
governanceuser-controllableinspectable
conflictllm-arbitratellm-arbitrate

Pros & cons

Anthropic Auto Dream

Pros: Bounded MEMORY.md size (200 lines) prevents context bloat; surgical updates preserve user-edited content; user-toggleable; complements rather than replaces CLAUDE.md.

Cons: Closed-source; opt-out mechanism not formally documented; consolidation logic opaque; no audit log for what was pruned/merged; only Claude Code (not Claude.ai web).

Official MCP Memory server

Pros: Reference implementation in the MCP spec; smallest possible memory primitive — useful for testing and lightweight cases.

Cons: Minimal feature set — no extraction, no governance, no retention policy; production deployments quickly outgrow it.

Rows last verified 2026-05-14 / 2026-05-14. Data is CC-BY-4.0 — see how to read this.