Open Source MCP Memory Server

Persistent memory for AI agents.

Built on Model Context Protocol.

Every session, your agents start from zero. HeurChain gives them structured memory that survives across sessions, models, and machines — with no prompt engineering required.

Deploy self-hosted → Start Workgroup — $49.99/mo

Performance

Built for speed. Designed for scale.

HeurChain is memory infrastructure — not a wrapper. Every number below is a design target, not a marketing claim.

Architecture

Three memory tiers that behave like memory, not a database

Memory stored yesterday behaves differently from memory stored six months ago — enforced structurally, not by policy.

ops/ 3× decay · HOT

Hot operational context

Current session data, active task state, real-time debugging trails. Fades fast by design — noise from yesterday shouldn't pollute today's focus.

Use for: Active task tracking, debug sessions, short-horizon agent state

notes/ 1× decay

Reference knowledge

Cross-session knowledge, summaries, learned facts. Standard ACT-R decay rate — information persists proportionally to how often it's accessed.

Use for: User preferences, project summaries, recurring facts, domain knowledge

self/ 0.1× decay

Agent identity

Persona definitions, behavioral constraints, long-term preferences. Near-permanent — decays at one-tenth the baseline rate. Core identity should outlast the session.

Use for: Agent persona, hard constraints, long-term user preferences, identity

Data Flow

How memory flows through HeurChain

Every write is indexed. Every read is fused. Every session starts with context.

Compatible Agents

Anthropic

Kimi

OpenAI

Nous

Three lines of code. Full memory stack.

→BM25 + Qdrant hybrid search, fused via Reciprocal Rank Fusion
→Session consolidation: ops/ compresses to 5–15 token cues automatically
→Working-group isolation at vault level, Redis namespace, and Qdrant collection
→MCP native — Claude Code, Cursor, Windsurf, any MCP client

View on GitHub Start Workgroup →

from heurchain import HeurChain

hc = HeurChain(
    url="http://localhost:3010",
    token="your-token"
)

# Store a memory
hc.add(
    "User prefers dark mode and speaks Spanish",
    user_id="user_123"
)

# Search memories
results = hc.search(
    "display preferences",
    user_id="user_123"
)

# Get proactive context at session start
context = hc.context(user_id="user_123")

{ "memories": [...], "decay_scores": [...], "relevance": 0.97 }

import { HeurChain } from "heurchain"

const hc = new HeurChain({
  url: "http://localhost:3010",
  token: "your-token",
})

// Store a memory
await hc.add(
  "User prefers dark mode and speaks Spanish",
  { userId: "user_123" }
)

// Search memories
const results = await hc.search(
  "display preferences",
  { userId: "user_123" }
)

// Get proactive context at session start
const context = await hc.context({ userId: "user_123" })

{ "memories": [...], "decay_scores": [...], "relevance": 0.97 }

# docker-compose.yml
services:
  heurchain:
    image: ghcr.io/peterjohannmedina/heurchain:latest
    ports:
      - "3010:3010"
    environment:
      - REDIS_URL=redis://redis:6379
      - QDRANT_URL=http://qdrant:6333
      - EMBED_URL=http://embedding:8080
      - BEARER_TOKEN=your-token
    depends_on:
      - redis
      - qdrant
      - embedding

  embedding:
    image: ghcr.io/peterjohannmedina/heurchain-embed:latest
    # BAAI/bge-m3 — GPU optional, CPU fallback included

  redis:
    image: redis:7-alpine
    volumes:
      - redis_data:/data

  qdrant:
    image: qdrant/qdrant:latest
    volumes:
      - qdrant_data:/qdrant/storage

volumes:
  redis_data:
  qdrant_data:

docker compose up -d # Running in <10 min

Pricing

Memory for your agents. Priced by the group.

One flat monthly rate per working group. Each group gets 10 million tokens of memory storage included. Most teams never come close.

Community

Self-Hosted

Free

MIT licensed. Run anywhere Docker runs. No account required.

All 3 memory tiers (ops/ notes/ self/)
ACT-R temporal decay engine
BM25 + Qdrant hybrid retrieval
Session consolidation
MCP server (stdio + HTTP)
Docker in under 10 minutes
MIT license

View self-hosted docs