Loopboard — Your Lovable agent finally knows where things stand
Persistent Memory for AI Coding Agents

The Persistent Memory Layer
for AI Coding Environments.

Stop letting your AI assistant lose context. Loopboard bridges the gap between developers and coding agents with a real-time, shared Kanban board. Zero context drift, zero regression.

60s

One-Click Setup

100%

Local & Private

Universal

Cursor & Claude Code

Workspace Board
Synced
In Progress
Dashboard UI Layout
feat/dashboard-ui+4 files
Done
REST API Layer
MCP Tunnel Established
AI Agent View
MCP Agent
Dev

What task should I focus on next?

I have loaded your board context. Proceeding to: Dashboard UI Layout.

mcp_task_get("t3")
The Context Cliff

AI agents fail when they lose track of the plan.

After several messages, chat histories bloat. AI assistants run out of context, forget recent technical decisions, and start overwriting functional code.

Standard Chat Environment
Memory Wiped
You:

Let's build the checkout page today.

AI Agent:

Sure! First, let's set up the database schema and initialize the OAuth authentication flow...

You:

Wait, we built and tested database schema and OAuth yesterday, remember?

AI Agent:

Apologies! My context window cleared. Could you paste the schema and auth code files so I can see what is implemented?

Loopboard Active Sync
Memory Active
You:

Let's build the checkout page today.

mcp_task_list(status: "done")→ OAuth 2.0 & DB Initialized
AI Agent:

Got it. I scanned your board: OAuth and database are marked Done. I will start writing the checkout integration code in `src/app/checkout` now.

Workspace: core-api
Linked

Saved Tokens

14,200

Sync Hits

152

Health

99.8%

In Progress
Stripe Checkout Integration
feat/checkoutActive Sync
Done
User auth setup
Developer Command Center

Total visibility into your AI's plan.

Loopboard acts as your control tower. Keep tabs on exactly what code paths your AI assistant is editing, which tasks are completed, and what is coming up next in the backlog.

  • Track progress in real-time across multiple active files and git branches.
  • Spot logical conflicts before they cause compile errors or regressions.
  • Perfect for handing off project state to a fresh developer or AI session.

Interactive Comparison

Maintain cognitive alignment.

See how an external Kanban board prevents context drift, reduces token spend, and keeps your AI agent on track.

Session Length: 48 Messages · 1,400 Prompt Tokens
Optimized Alignment
You:

"Let's work on the Stripe payment integration now."

Tool Call: mcp_task_recall()Retrieving board snapshot. Verified State: SQLite Schema (Done), Auth Router (Done), Checkout Integration (To Do).
AI:

Got it. I loaded your Loopboard sprint context via MCP. Database schema and authentication system are fully completed. I'll proceed directly to coding the Stripe checkout integration files. Should we create it under src/components/billing?

Board Memory — Injected Context
[Brief] A SaaS for book clubs. Users sell used books, join clubs, and track reading.
[Vocabulary]
"Workspace" = user's book collection space (NEVER call it "team")
"Club" = paid or free book club (NEVER call it "group")
"Shelf" = inventory of books for sale (NEVER call it "store")
[Rules]
• Trial users cannot invite teammates
• Paid users can invite up to 5
• Admins can invite unlimited
[Shipped Features]
✓ Auth (OAuth + email) — DONE Mar 1
✓ Dashboard UI — DONE Mar 3
✓ Pricing page — DONE Mar 5

Board Memory

Your AI never forgets. You set the rules.

Persistent project context that lives on the board — not in the chat. Lock a brief, define your vocabulary, and capture rules once. Your AI reads them on every session.

Locked Project Brief

Write a one-paragraph brief once. Loopboard locks it on save and feeds it verbatim into every MCP session. The AI can't drift from your core mission.

Vocabulary Lock

Tell the AI that "Workspace" never means "team", that "Club" never means "group", that "Shelf" never means "store". Term locks prevent costly renames later.

Rules & Constraints

Capture business rules you keep re-explaining: "Trial users cannot invite teammates", "Always use Tailwind, never inline styles". Categorized for easy scanning.

Auto-Synced Shipped Features

Tasks in the Done column auto-flow into AI context as a 'shipped features' digest. The agent never asks you to re-build something you already shipped.

Board Memory
Project Brief
LOCKED
A SaaS for book clubs. Users sell used books, join clubs, and track reading.
Vocabulary Lock3

"Workspace" = user's book collection space

NEVER call it "team"

"Club" = paid or free book club

NEVER call it "group"

"Shelf" = inventory of books for sale

NEVER call it "store"

Rules & Constraints3
AUTH

Trial users cannot invite teammates

AUTH

Paid users can invite up to 5

AUTH

Admins can invite unlimited

Tasks in the Done column are auto-synced to AI context
See it live on your board

Memory is a first-class MCP resource. Read with memory_get, add to it with memory_add_vocabulary and memory_add_rule.

Transparent, developer-first rates

Flexible plans for any scale

Free for hobby side-projects. Scalable upgrades for commercial shipping.

Hobby
$0/ free forever

Perfect for side projects and local AI testing.

  • 3 active boards
  • 100 tasks limits
  • Unlimited MCP hits
Get Started
Most Popular
Pro
$9/ month

For professional developers shipping software with AI.

  • 10 active boards
  • 500 tasks limits
  • Vector semantic recall
Start 14-day free trial
Team
$29/ month

For collaborative teams and startup squads.

  • Unlimited boards
  • 2,000 tasks limits
  • Full semantic indexing
Start 14-day free trial
Enterprise
Custom

For organizations requiring strict data security policies.

  • Unlimited tasks
  • SSO & SAML support
  • Dedicated hosting support
Contact Sales

See how it works

Watch your board stay in sync while you build — no manual updates needed.

AI Developer AgentMCP Persistent Memory
Hey! What are the remaining tasks?
Interactive Simulator Controls:
Workspace: my-saas-app
To Do
2
#billingt1

Stripe billing flow

feat/stripe-billing2h ago
#emailt2

Email notifications

feat/email-notif5h ago
In Progress
2
#uit3

Dashboard UI

feat/dashboard-ui+4 files10m ago
#backendt4

REST API layer

feat/api-layer+8 files3m ago
Done
1
#backendt5

Database schema

feat/db-init+3 files1d ago

Orchestrate AI workflows with precision.

Loopboard serves as the centralized source of truth for you and your AI agent, making session handoffs seamless.

Real-Time Context Synchronization

As your AI assistant generates code or refactors routes, Loopboard updates task cards on the fly. No manual tracking, no stale tasks, and no out-of-sync backlogs.

Active Listening Tunnel

Local-First Privacy

Maintain full ownership of your plans. Run Loopboard completely offline as a local CLI tool, or use encrypted tokens to sync safely in our secure cloud.

Zero-knowledge architecture

Semantic Task Recall

When context gets dense, Loopboard uses vector search to retrieve relevant tasks, requirements, and structural schemas. Your AI loads only what is needed.

Context compression engine

Universal MCP Support

Connect to any AI coding workspace in seconds. Loopboard exposes a standard Model Context Protocol (MCP) server compatible with Claude Desktop, Cursor, Lovable, Cline, Windsurf, and custom LLM scripts.

npx @cwim/kanbanGitHub Repository

Board Memory — your AI never forgets

Lock a project brief, pin vocabulary, capture rules, and let shipped features auto-sync. Persistent context that lives on the board and is read on every MCP session.

Brief
Locked on first save
Vocabulary
term → meaning → never
Rules
categorized constraints
Shipped
auto-synced from Done
Model Context Protocol APIs

How Loopboard communicates with AI.

Loopboard exposes a standard MCP interface that coding assistants query directly. Click any tool function below to see the exact payload sent by your agent.

mcp_tool_inspector: task_create()
Functional Description

Creates a brand new task card on the board with tags, description, and status column.

Payload Schema (Request)
{
  "title": "Implement OAuth Auth",
  "status": "todo",
  "tags": ["auth", "backend"]
}
Callback Context (Response)
{
  "success": true,
  "created_id": "t6",
  "title": "Implement OAuth Auth",
  "status": "todo"
}

Connect in 60 seconds.

Standard Model Context Protocol (MCP) configuration. Integrate with any compatible agent environment.

STEP 01

Initialize Your Board

Sign in to your Loopboard dashboard and create a workspace board for your active development project.

STEP 02

Copy Connection URL

We generate a secure, authenticated connection URL unique to your workspace board.

https://loopboard.dev/api/mcp
STEP 03

Link to AI Client

Paste the URL into your AI editor settings (Cursor, Claude, Cline, or Lovable). Context compiles immediately.

Frequently Asked Questions

Everything you need to know about Loopboard.

Keep your AI agent aligned.
Start building without context drift.

Free tier provides 3 workspace boards. Connection takes under 60 seconds with standard MCP servers.