Skip to main content
1

Generate an API key

Go to Settings > API keys in your Kommit dashboard. Click Generate key and copy the key (prefixed km_).
API keys are shown once. Store them securely — you can’t retrieve them later.
2

Configure your AI tool

Add the Kommit MCP server to your tool’s configuration:
Add to .claude/settings.json in your project root (or ~/.claude/settings.json for global):
{
  "mcpServers": {
    "kommit": {
      "url": "https://mcp.getkommit.ai/sse?token=km_your_api_key"
    }
  }
}
3

Verify the connection

Start a new session in your AI tool and ask it something about your project. It should call search_memory or get_project_overview automatically.If the tool reports a connection error, verify:
  • The API key is correct and not expired
  • The key has the right project scope
  • Your network allows outbound HTTPS to mcp.getkommit.ai

​
Available tools

Once connected, your AI tool can use these Kommit tools:
Search project memories by semantic similarity.Parameters:
  • query (string, required) — what you’re looking for
  • project_id (string, optional) — scope to a specific project
  • category (string, optional) — filter by decision, learning, progress, or preference
Store a new memory (decision, learning, or preference).Parameters:
  • content (string, required) — the memory content
  • category (string, optional) — decision, learning, progress, or preference
  • project_id (string, required) — which project this belongs to
Get a full snapshot of a project — description, target audience, tech stack, features, and memory stats.Parameters:
  • project_id (string, required) — the project ID
Retrieve the full project specification including all canvas nodes and their structured data.Parameters:
  • project_id (string, required) — the project ID
Get the project’s data model definitions.Parameters:
  • project_id (string, required) — the project ID
Get details about a specific feature node.Parameters:
  • feature_id (string, required) — the feature node ID
List all projects in the organization.No required parameters.

​
Rate limits

API keys are rate-limited to 1,000 requests per hour. MCP tool calls count toward this limit. If you hit the limit, the server returns a 429 status and the AI tool will see an error.