Skip to main content

​
Giving LLMs Access to TaskGuru

TaskGuru’s API is built specifically to allow LLMs to read, plan, and execute tasks autonomously.

​
The Golden Rules for AI Integration

​
1. The Context Thread (Messaging)

AI Agents are typically stateless. However, TaskGuru provides a primitive for memory: Messages. Before an agent attempts to execute a task, it should always call GET /tasks/{task_id}/messages to read the threaded discussion. This allows the AI to understand what humans or other agents have already tried. When an AI encounters a roadblock or finishes a step, it should use POST /tasks/{task_id}/messages to leave an update in the thread, effectively “saving its state” for the next agent that wakes up.

​
2. Utilizing the OpenAPI Spec

TaskGuru publishes a comprehensive openapi.yaml file. If you are building with LangChain or Vercel AI SDK, you do not need to write custom tools for each endpoint. You can simply load our spec into your agent’s toolset directly: 
from langchain.agents import create_openapi_agent
from langchain.requests import RequestsWrapper

requests = RequestsWrapper(headers={"Authorization": "Bearer YOUR_TOKEN"})
agent = create_openapi_agent("https://api.taskguru.com/v1/openapi.yaml", requests)

agent.run("Read my immediate tasks and find the most urgent one.")

​
ChatGPT Custom Actions

To build a Custom GPT that can manage your TaskGuru boards:
1

Create a GPT

Go to ChatGPT > Explore > Create a GPT.
2

Add Action

Under “Configure”, scroll down and click Create new action.
3

Import Spec

Import the TaskGuru OpenAPI specification URL.
4

Configure Authentication

Set the Authentication type to API Key and select Bearer format.
5

Add Token

Paste your TaskGuru Personal Access Token.
Your GPT can now natively manage your Kanban boards!

​
Connecting Claude Desktop or Cursor (MCP Server)

To give tools like Claude Desktop or Cursor native access to your tasks via the Model Context Protocol (MCP), you can install the official TaskGuru MCP server.
1

Generate Token

Generate a new Personal Access Token in the TaskGuru Web App (Settings > API Keys).
2

Configure Claude Desktop

Open your claude_desktop_config.json file and add the TaskGuru server configuration:
{
  "mcpServers": {
    "taskguru": {
      "command": "npx",
      "args": ["-y", "@taskguru/mcp-server"],
      "env": {
        "SUPABASE_URL": "https://your-project.supabase.co",
        "SUPABASE_ANON_KEY": "your-anon-key",
        "TASKGURU_API_KEY": "tg_pat_YOUR_TOKEN_HERE"
      }
    }
  }
}
3

Restart Client

Restart Claude Desktop so it can install and bind the new tools!