Add Banyan as an MCP server to your AI assistant in under a minute.
Claude.ai (web) and Claude Desktop both use the same connector screen. The setup below works for either one.
Banyan and enter the URL:https://api.banyan.vamitra.com/mcp/auth
What to say:
"Check in with Banyan and show me what's there"
"I want to research [topic] — create a knowledge tree for it"
"What ideas are on my idea board?"
Two ways to connect. The plugin (recommended) gets you tools + slash commands + session hooks + the operating-principles skill in one install. Manual MCP gets you just the tools.
/plugin marketplace add badriram/banyan-claude-plugin
/plugin install banyan@banyan-plugins
What you get:
claude mcp add needed/banyan:checkin, /banyan:capture, /banyan:status, /banyan:goalclaude mcp add --transport http banyan https://api.banyan.vamitra.com/mcp/auth
What to say:
"Check in with Banyan, look at the idea board for open tasks"
"What shipped in the last two days?"
"Add this design decision to the architecture docs"
Banyan supports any MCP client via OAuth 2.1 or API key authentication.
MCP endpoint: https://api.banyan.vamitra.com/mcp/auth
Discovery: https://api.banyan.vamitra.com/.well-known/oauth-authorization-server
MCP endpoint: https://api.banyan.vamitra.com/mcp
Header: x-api-key: YOUR_KEY
Contact banyan-support@vamitra.com for an API key.
A second MCP endpoint exposes operator tools — forest / grove creation, member management, list_forests / list_groves, prune, lint, discard_capture. Most clients don't need it; only attach when you're running org-level admin actions.
MCP endpoint: https://api.banyan.vamitra.com/mcp/admin
Auth: OAuth 2.1 (Cognito-only — no API key). Discovery URL same as above.
You don't need to know tool names. Just describe what you want — your AI assistant figures out which Banyan tools to use.
Research
"I want to explore whether GLP-1 drugs improve heart outcomes — set up a research tree with branches for literature, mechanisms, and trials"
Product planning
"Create a product planning trunk for Q3. I need branches for feature ideas, design decisions, and shipped work"
Capture & recall
"That insight about caching strategies is good — save it to Banyan so I can find it later"
Team collaboration
"Invite jordan@acme.com to my research forest as a reviewer. Set the trunk so contributions need human approval"
Cross-session memory
"Pull up my architecture notes from last week — I need to pick up where I left off"
Agent coordination
"Check in as researcher on the AI comparison trunk, see what tasks are assigned to me"
Banyan stores your knowledge as trees — trunks (ideas), branches (angles), leaves (content), and connections between them. Read the full guide to learn more.