MEMORY LAYER FOR AI AGENTS

YOUR AGENTS
FINALLY REMEMBER.

One shared context layer for sessions, docs, facts, and project history across Cursor, Claude Code, and Windsurf.

$npx memoire-ai init
CURSORCLAUDE CODEWINDSURFCODEXAMP
// CONNECTS WITH YOUR FAVORITE TOOLS
CursorCursor
ClaudeClaude
WindsurfWindsurf
GitHubGitHub
SlackSlack
NotionNotion
LinearLinear
OpenAIOpenAI
// SEE IT IN ACTION

MEMOIRE INSIDE YOUR TOOLS

Shared memory, searched and assembled into prompts. Every session starts where the last one ended.

auth.ts — acme-api
TSauth.ts
TSproxy.ts
1
2
3
4
5
6
7
8
9
10
11
12
import { verify, sign } from 'jsonwebtoken'
import { LRUCache } from 'lru-cache'
// Session cache — switched from Redis (see memoire)
const cache = new LRUCache({ max: 500 })
export async function authMiddleware(req) {
const token = req.headers['authorization']
const session = cache.get(token)
if (session) return session
const payload = verify(token, process.env.JWT_SECRET)
// HOW IT WORKS

THREE STEPS TO
PERSISTENT CONTEXT

Memoire captures everything from your coding sessions, enriches it with facts and version-matched docs, then assembles it into a token-budgeted prompt for the next session.

>capture → enrich → retrieve
Session MemoryCaptures every decision
// FEATURES

WHAT MEMOIRE REMEMBERS

Sessions, facts, docs, and project activity — captured, enriched, and retrieved.

01

SESSION MEMORY

Capture prompts, decisions, tool use, and summaries. Next session starts with full context — no re-explaining.

02

CROSS-CLIENT SHARING

Cursor, Claude Code, Windsurf, and Codex share one memory backend. Switch tools without losing context.

03

VERSION-AWARE DOCS

Indexes your package.json dependencies and retrieves documentation matching your exact versions. No stale training data.

04

PROMPT ASSEMBLY

Search results, timelines, facts, and docs — ranked and assembled into a token-budgeted prompt block.

05

SPO EXTRACTION

Automated subject-predicate-object fact extraction with contradiction detection. Old facts get superseded, not duplicated.

5 CORE CAPABILITIES
// CONNECTORS

CONNECTED TO YOUR STACK

Sync context from the tools your team already uses. Every connector feeds into Memoire's shared memory.

GitHub
GITHUB
Slack
SLACK
M
MEMOIRE
Notion
NOTION
Linear
LINEAR
// ENTERPRISE

BUILT FOR TEAMS

Security, access control, and deployment flexibility for engineering organizations that need their memory layer to be as reliable as their CI pipeline.

BOOK A DEMO
01

RBAC

Four roles — org_admin, project_admin, member, viewer. Granular API key scoping per project and environment.

02

ENCRYPTED CONNECTORS

Application-level encryption for all OAuth tokens. Slack, Notion, GitHub, and Linear credentials never touch disk unencrypted.

03

AUDIT TRAIL

Full memory_history table with immutable event log. Every change, every access, every query — recorded and queryable.

04

SELF-HOST

Run Memoire on your own infrastructure. Docker Compose, AWS ECS, or bare metal. Your data never leaves your network.

05

PRIVACY STRIPPING

Collector pipeline strips PII before storage. Configurable capture policies: standard, lean, or debug.

06

SSO-READY

Supabase Auth with magic link and OAuth today. Enterprise SSO via SAML on the roadmap.

AGPL-3.0POSTGRESPGVECTOR
// PRICING

TRANSPARENT. NO SURPRISES.

Open source core. Hosted option for teams that don't want to manage infrastructure.

FREE / OPEN SOURCE
$0
self-hosted, forever
  • >Core memory system
  • >MCP server
  • >CLI tools
  • >Community support
INSTALL
PRO
$29
per seat / month
  • >Everything in Free
  • >Cloud hosted
  • >Team collaboration
  • >All connectors
  • >Priority support
START TRIAL
ENTERPRISE
CUSTOM
contact sales
  • >Everything in Pro
  • >SSO / SAML
  • >Audit logging
  • >SLA guarantee
  • >Dedicated support
BOOK DEMO
// GET STARTED

STOP RE-EXPLAINING.
START BUILDING.

Install Memoire once, connect your project, and let each new session start with shared context instead of a blank prompt.