Skip to main content
The Model Context Protocol (MCP) is an open standard for connecting AI tools to external data sources. Kommit implements an MCP server that exposes your project specs, memory, and data schemas to AI coding tools.

​
Why MCP matters

Without MCP, every AI coding session starts from scratch. The AI reads your code but doesn’t know:
  • What decisions you’ve made and why
  • What the product requirements are
  • How the data model is designed
  • What conventions your team follows
Kommit solves this by serving as the memory layer. When your AI tool sends a query, Kommit searches your project memory and returns relevant context.

​
Available MCP tools

When connected, AI tools can call these Kommit tools:
ToolDescription
search_memorySearch project memories by semantic similarity
add_memoryStore a new decision, learning, or preference
get_project_overviewGet a full snapshot of the project — description, features, tech stack
get_specRetrieve the project specification (canvas nodes and edges)
get_data_schemaGet the project’s data model definitions
get_featureGet details about a specific feature node
get_tech_stackGet the project’s technology choices
list_projectsList all projects in the organization

​
How it works

  1. You generate an API key in Kommit (prefixed km_)
  2. You add the Kommit MCP server URL to your AI tool’s configuration
  3. When the AI needs context, it calls Kommit’s MCP tools
  4. Kommit authenticates the request, resolves the project, and returns relevant data
See MCP setup for configuration instructions.