Notion Career Sync: 1-Click AI Job Tracker via Notion MCP

This is a submission for the Notion MCP Challenge

What I Built

Hunting for SDE intern roles and full-stack positions usually means juggling dozens of open tabs. Whether it's LinkedIn, Wellfound, or a company's custom career portal, manually copying and pasting the company name, role, salary, and tech stack into a tracking spreadsheet is a tedious, friction-heavy process.

I wanted to build a "human-in-the-loop" system to completely automate this workflow.

Notion Career Sync is an intelligent Chrome Extension paired with a local MCP server. With a single click on any job posting, the extension captures the page context and sends it to a local Node.js backend. From there, an AI agent (powered by Groq and Llama 3.3 70B) parses the unstructured text to extract the core job details. Finally, it uses the Notion MCP to instantly create a highly structured, categorized row in my Notion Job Tracker database. No more copy-pasting—just click, extract, and track.

Video Demo

Show us the code

Notion Career Sync GitHub Repository

How I Used Notion MCP

To make this system robust, I needed a way for my local AI agent to seamlessly write to my workspace without writing brittle, custom API wrappers for Notion.

I integrated the @modelcontextprotocol/sdk directly into my Express backend. Here is the workflow it unlocks:

1. The Transport: The backend uses StdioClientTransport to spawn and communicate with the official @modelcontextprotocol/server-notion as a child process.
2. The Handshake: Groq's LLM processes the messy DOM text from the Chrome extension and outputs a clean JSON object containing the company, role, location, salary, and techStack.
3. The Action: Instead of making standard REST calls, the server acts as an MCP client, dynamically calling the Notion tool to append a new page to my Job Tracker database. It perfectly maps the AI's extracted tech stack array to my Notion multi-select properties and sets the default status to "Applied".

Using the Notion MCP transformed what would have been a complex integration headache into a clean, standardized bridge between my local LLM and my Notion workspace. It gives the AI "hands" to organize my career hunt automatically!

Comments

[Hamza KONTE]

Cool project! MCP integration for productivity tools is really where agentic workflows get practical. If you want to add prompt structuring to your Claude workflow, flompt also has an MCP server: claude mcp add flompt https://flompt.dev/mcp/ -- your agents can call decompose_prompt and compile_prompt natively. Pairs well with Notion MCP for building well-structured task descriptions before they hit your tracker. Free, open-source. A star on GitHub means a lot: github.com/Nyrok/flompt

[Swift]

This is awesome. Thanks for posting on DEV!