Kankali Context
Open Source Loading…
100% Open Source · Self-Custodial AI Context Protocol

One Persistent Memory.
Every AI Assistant.

Eliminate context fragmentation across Claude, Cursor, ChatGPT, and Grok. Kankali provides a transparent, stateless memory layer that stores project architecture, memory chunks, and SDLC docs directly in your private Google Drive AppData and GitHub Repository.

Explore Vault UI Source Code
AES-256-GCM ENCRYPTED
OAUTH 2.1 + PKCE (RFC 7636)
GOOGLE DRIVE APPDATA SANDBOX
100% OWNED IN YOUR GITHUB

Unified Master MCP Endpoint

Connect Claude, Cursor, ChatGPT, or Grok to your persistent memory vault in seconds with zero login friction using signed Master Auth URLs.

  • Self-Custodial: Memory files live in your private Drive or GitHub, never on central disks.
  • Anchor Chunking: Saves 95% tokens by reading specific heading blocks (read_outline).
  • Stateless & Auditable: 100% open source under MIT License. No AI model training on your data.
MASTER MCP CONNECTOR URLACTIVE (V2.0)
https://kankali-context.vercel.app/mcp/master
Copy your Zero-Login signed Master Auth URL from Settings.

How Kankali Operates

A transparent, four-step process to persistent AI memory.

1

Sign In

Authenticate with Google OAuth to enable your private Google Drive AppData vault.

2

Connect Storage

Optionally link your private GitHub repository for team versioning and git commits.

3

Attach MCP Connector

Paste your Master Auth URL into Claude, Cursor, ChatGPT, or Grok as a Custom MCP server.

4

Persistent Memory

Context flows seamlessly across all sessions with anchor-based micro-chunk retrieval.

Zero AI Training & Self-Custodial Isolation

Kankali acts purely as a stateless transport layer. We do not inspect, log, or train on your contextual data. Your memory vault resides exclusively in your own Google Drive AppData sandbox or your designated GitHub repository.

REVIEW SECURITY ARCHITECTURE & AUDIT →