Skip to content
knktx — unified context service for AI agentsknktx
Early Access

The Unified
Context Service
for AI Agents.

Instantly bootstrap any AI agent from anywhere. Sync workflows, rules, and live project context across machines so your agents never lose progress, and you maintain total oversight.

>uvx knktx init
SaaS Authentication Setup Live Sync

Track agent progress in real-time. The board auto-updates as agents execute the workflows.

Configure OAuth Providers
Inbox
Implement JWT Middleware
Inbox
Setup Role-Based Access
Inbox

Works with any MCP agent

Claude Code
Gemini CLI
Codex CLI

Dual-Layer Architecture

Designed simultaneously for human comprehension and machine execution. The ultimate bridge between intent and automation.

Visual Oversight for Humans

Manage context and see your work visually. Agents handle the heavy lifting of creating projects, moving tasks, and drafting plans via MCP based on your commands.

MCP for Agents

An autonomous execution layer. Agents read your context, update statuses, and query project memory natively via Model Context Protocol.

Universal Context

State follows you everywhere. Works flawlessly across Claude, Gemini, and Codex CLI. No more disconnected chats or lost context.

Built for Resilience

Features engineered to prevent context loss, ensuring your AI workflows remain robust and continuous, no matter where you work.

  • 65+ MCP Tools
  • 10 Entity Types
  • 3 Agents Compatible

Intelligent Tracking

Notes, plans, and documents automatically track their own implementation progress across linked task cards.

Full CRUD API

Every entity — projects, boards, cards, documents, workflows, rules, skills, notes, plugins, and MCP configs — has complete create, read, update, and delete coverage via MCP.

Workflow Sync CLI

Run knktx init to bootstrap or knktx update to sync. Rules, workflows, MCPs, skills, and plugins stay in sync across all your machines.

Real-time Kanban

WebSocket-powered board updates. See agent progress live as cards move through columns — no refresh needed.

Persistent Memory

Project context, rules, and preferences are durably stored, meaning your agents never start from a blank slate.

Activity Attribution

Every action is tagged to the specific agent model that performed it. Full audit trail across Claude, Gemini, and Codex.

Public Gallery

Browse, preview, and fork community rules, workflows, and skills. One-click fork with full attribution chain.

Injection Modes

Choose full or soft injection. Full injects rules and complete workflows for strict adherence — heavier on tokens. Soft injects rules plus workflow IDs and descriptions so agents match and pull only what they need.

Simple, Transparent Pricing

Currently invite-only. Request access to get started.

Developer

Free

Perfect for solo developers and personal projects.

  • 1 Project
  • 3 Boards per project
  • Unlimited Cards
  • Cloud Context: Rules, Workflows, Notes, Documents & Plans
  • Public Gallery: Browse & Fork Community Content
Request Access
Recommended

Pro

Free

Unlimited access during early launch. $20/mo after general availability.

  • Everything in Developer
  • Unlimited Projects
  • Unlimited Boards
Request Access

Team

Coming Soon

For engineering teams collaborating with AI.

  • Everything in Pro
  • Shared Team Context
  • Role-Based Access Control
  • Organization Workflows
Join Waitlist

Frequently Asked Questions

Everything you need to know about knktx.

knktx is a centralized context service for AI agents. It stores your project plans, tasks, workflows, rules, skills, and MCP configs in one place — accessible to any AI agent via MCP. It also injects rules and workflows directly into your agent's instructions file (CLAUDE.md, GEMINI.md, AGENTS.md) so agents follow your playbook from the moment they start. Instead of scattering context across local files, agents read and write to knktx directly.

Give your agents memory.

Stop re-explaining context. Start shipping with agents that know your project, follow your rules, and pick up where they left off.