MCP 服务器

Dense Mem

About Self-hosted AI agent memory server with MCP, evidence provenance, typed claims, conflict detection, embeddings, recall, PostgreSQL, and Neo4j.

README

<h1 align="center">Dense-Mem</h1>

<a href="README.md">English</a> · <a href="README.zh-CN.md">简体中文</a>

Self-hosted memory for AI agents that preserves evidence, detects conflicts, and never silently rewrites facts.

Create a temporary isolated team and test Dense-Mem before self-hosting.

Dense-Mem gives MCP clients a durable memory layer with provenance, typed claims and facts, verification gates, server-side embeddings, recall, team isolation, REST/OpenAPI, and a token-protected control portal. The host LLM owns conversation and judgment; Dense-Mem owns durable memory state and returns structured outcomes the host can explain to users.

Under the hood, Dense-Mem is a standalone HTTP MCP memory server. HTTP MCP is the v1 supported MCP transport and is served at /mcp from the main HTTP process.

Dense-Mem is part of the research preprint Governed Enterprise AI Memory Beyond RAG: From Vector Retrieval to Permissioned Knowledge Graphs.

Project Intro

<a href="https://cdn.markhuang.ai/videos/dense-mem/intro.mp4" target="_blank" rel="noopener noreferrer">Watch the Dense-Mem intro video</a>

Try the Hosted Demo

Create a temporary isolated team at https://demo-dense-mem.markhuang.ai and test Dense-Mem before self-hosting.

Why Dense-Mem?

AI agents need memory that can be trusted later, not only text that can be retrieved later.

Evidence is first-class. Memories start as source fragments before they become claims or facts.
Facts pass through typed claims, verification, and promotion gates.
Comparable conflicts become clarifications[]; Dense-Mem does not silently overwrite active facts.
The host LLM stays responsible for extracting candidates and asking the user questions. Dense-Mem stays responsible for durable state, gates, audit metadata, and recall.
Operators keep control of storage, team/profile isolation, API keys, and data egress boundaries.

60-Second Quickstart

Download the base local-only compose example and env template, set the required secrets, and start Dense-Mem:

mkdir dense-mem-local
cd dense-mem-local

curl -fsSLo docker-compose.yml \
  https://raw.githubusercontent.com/markhuangai/dense-mem/main/examples/docker-compose.base.yml
curl -fsSLo .env.example \
  https://raw.githubusercontent.com/markhuangai/dense-mem/main/examples/.env.example

cp .env.example .env
# Fill in POSTGRES_PASSWORD, NEO4J_PASSWORD, CONTROL_PORTAL_TOKEN, and AI_API_KEY.
${EDITOR:-vi} .env

docker compose up -d
docker compose exec server /app/provision-team --name "primary-memory"

The base compose example provisions Postgres, neo4j:5.26-community with the Neo4j Graph Data Science plugin, and the Dense-Mem server. It exposes only local host ports:

MCP/API:        http://127.0.0.1:8080/mcp
User portal:    http://127.0.0.1:8080/ui
Control portal: http://127.0.0.1:8090/

Cold image pulls can take longer than 60 seconds. Redis and public HTTPS are intentionally omitted from the base example; use the expert example when you need those deployment options.

The server requires a complete embedding configuration at startup: AI_API_URL, AI_API_KEY, AI_API_EMBEDDING_MODEL, and AI_API_EMBEDDING_DIMENSIONS. The compose examples provide OpenAI defaults for the URL, model, and dimensions (https://api.openai.com/v1, text-embedding-3-small, 1536), so the minimal local setup only needs you to fill in AI_API_KEY. Override those values together when using a different embedding provider or model.

Telemetry Overlay

Prometheus telemetry is optional and off by default. To collect usage, performance, verifier token, embedding token, recall, and promotion metrics for the /ui app and control portal dashboards, run the base stack with the telemetry overlay:

curl -fsSLo prometheus.yml \
  https://raw.githubusercontent.com/markhuangai/dense-mem/main/examples/prometheus.yml
curl -fsSLo docker-compose.telemetry.yml \
  https://raw.githubusercontent.com/markhuangai/dense-mem/main/examples/docker-compose.telemetry.yml

export TELEMETRY_SCRAPE_TOKEN="$(openssl rand -hex 32)"
docker compose -f docker-compose.yml -f docker-compose.telemetry.yml up -d

The overlay starts Prometheus on 127.0.0.1:9090, retains 30 days of samples, passes TELEMETRY_SCRAPE_TOKEN to Prometheus as a scrape secret, and points Dense-Mem at http://prometheus:9090 for telemetry queries. It also sets TELEMETRY_PROMETHEUS_JOB=dense-mem so dashboards query only the dense-mem scrape job when Prometheus is shared.

Online recall-quality cards use densemem_recall_feedback_total and densemem_recall_feedback_quality_score. They stay at zero until RECALL_FEEDBACK_ENABLED=true registers the submit_recall_feedback tool and a host LLM submits compact feedback for recall_memory results. Normal production recall traffic still contributes request volume, result count, and latency.

For the disposable demo image, keep the control portal disabled and use the demo telemetry overlay instead:

curl -fsSLo prometheus.demo.yml \
  https://raw.githubusercontent.com/markhuangai/dense-mem/main/examples/prometheus.demo.yml
curl -fsSLo docker-compose.demo.telemetry.yml \
  https://raw.githubusercontent.com/markhuangai/dense-mem/main/examples/docker-compose.demo.telemetry.yml

export TELEMETRY_SCRAPE_TOKEN="$(openssl rand -hex 32)"
docker compose -f docker-compose.yml -f docker-compose.demo.telemetry.yml up -d

The demo overlay scrapes the demo service at demo:8091 on the private Compose network and sets TELEMETRY_PROMETHEUS_JOB=dense-mem-demo. Do not publish that metrics listener publicly.

Compare

Capability	Dense-Mem	File memory	Vector DB	Generic MCP memory
Evidence provenance	Source fragments are stored before claims or facts	Usually absent or informal	Stores chunks, not truth history	Varies by implementation
Fact changes	Verification gates and promotion rules	Manual edits	Similarity updates can obscure history	Often tool-specific
Conflict handling	Comparable conflicts return clarification tasks	Caller must notice	Similar vectors do not mean contradiction	Usually caller-managed
Recall	Facts, claims, fragments, contradictions, and clarifications	Text search	Vector similarity	Varies
Agent boundary	Host LLM judges; Dense-Mem stores and enforces	Blurred	Retrieval only	Often blurred
Operations	Teams, profiles, API keys, audit metadata, REST, OpenAPI, MCP	Minimal	Database operations	Varies

Redis is optional for single-node deployments and required for multi-instance deployments.

Documentation

The README is the product overview. The full user documentation lives in the Dense-Mem wiki:

Goal	Wiki page
Run Dense-Mem locally	Quick Start
Use memory day to day	Using Dense-Mem
Configure providers, Redis, and Traefik	Configuration
Understand the system design	Architecture
Review API and operations details	Technical Reference

Responsibility Boundary

Area	Dense-Mem owns	Host LLM owns
Memory writes	Evidence fragments, typed claims, verification, gates, promotion	Extracting candidate memories from chat text
Embeddings	Fragment embeddings and recall-query embeddings through the configured provider	No vectors for normal writes or recall
Retrieval	Facts, validated claims, fragments, contradictions, clarification tasks	Choosing what to ask or cite in the conversation
Truth changes	Comparable-conflict detection, confirmation-driven supersession	Asking the user which uncertain memory is correct
Operations	Teams, named profiles, API keys, audit metadata, control portal	Client-side MCP configuration

Dense-Mem is not an agent brain, planner, or external truth arbiter. It stores memory, applies explicit gates, and returns structured outcomes.

Memory Workflow

Tool	Purpose
`remember`	Normal chat-session memory insertion. Saves evidence, creates typed claims, verifies, promotes when gates pass, and returns structured outcomes.
`import_memories`	Ingests summarized historical conversations. By default it records evidence and validated claims without auto-promotion.
`recall_memory`	Retrieves facts, validated claims, fragments, and `clarifications[]` for the authenticated team.
`trace_memory`	Expands one fact or claim into bounded evidence, promotion lineage, contradictions, and supersession links.
`assemble_context`	Builds a bounded prompt-ready context block plus structured facts, claims, fragments, and clarifications.
`reflect_memories`	Reviews active facts, candidate or disputed claims, contradictions, stale memories, and clarification needs.
`confirm_memory`	Applies the user's answer to a clarification task, either accepting a claim and superseding comparable active facts or keeping/rejecting it.

Low-level tools remain available for advanced callers: save_memory, post_claim, verify_claim, promote_claim, search tools, graph query tools, community tools, and retraction tools.

Memory moves through this path:

source fragment -> typed claim -> verification -> promotion gate -> active fact
                                                   |
                                                   v
                                            clarification task

Comparable conflicts are not resolved silently. Dense-Mem returns clarifications[], and the host LLM asks the user which memory is correct. After the user answers, the host calls confirm_memory.

Data Egress

Dense-Mem forwards fragment text and recall queries to the configured embedding provider. Claim verification can send candidate claims and supporting evidence to the configured verifier provider. Self-hosted providers keep that traffic inside your boundary; hosted providers do not. See the wiki Configuration and Technical Reference for provider settings and egress details.

Embedding Model Consistency

Dense-Mem owns embeddings for normal writes and recall. It checks the stored embedding model and dimension on startup so vectors from incompatible models are not mixed silently. Rotation requires re-embedding or rebuilding vector indexes; the step-by-step process belongs in the wiki Configuration.

Tool Discoverability

Dense-Mem exposes three discoverability surfaces backed by one registry:

Surface	Path	Purpose
Tool catalog	`GET /api/v1/tools`	Runtime tool discovery
Runtime OpenAPI	`GET /api/v1/openapi.json`	Agents, codegen, integrations
MCP Streamable HTTP	`POST /mcp`, `GET /mcp`	MCP clients over the main HTTP service

The full route list and client examples live in the wiki Technical Reference and Quick Start.

Design Notes

License

Apache-2.0