Skip to content

MVP Portal Automation — Contribution Entry AutoPilot

2025 | Python, Playwright MCP, Azure OpenAI, Agno | Private Utility Project

Problem / Motivation

Submitting contributions to the MVP Portal is repetitive, error-prone, and time-consuming. Each activity requires typing or pasting the same fields into forms, often leading to formatting errors, character overflows, or missing private descriptions.

This project reduces those friction points by: - Accepting structured input (MVP-PENDING.md) as the single source of truth. - Enforcing validation rules (character limits, required fields, dates). - Generating missing private descriptions automatically. - Using a browser automation agent (Playwright MCP) to fill the form exactly as a human would. - Pausing for human review before final save.

Overview

The solution consists of two key parts:

  1. MVP Portal Automation: Python-based workflow that parses contribution data from markdown, enriches it using AI, validates fields, and automates browser interaction.

  2. MVP Activity Copilot (Proof of Concept): Experimental agent built using Playwright MCP and Agno. It takes snapshots of the MVP Portal contribution form, fills in values dynamically, and continues based on AI prompting.

The Copilot is purely experimental, not supported by Microsoft or the MVP Program, and intended only for learning and personal workflow automation.

Tech Stack & Frameworks

  • Languages / Frameworks: Python, Node.js, PowerShell
  • Automation: Playwright MCP (browser-driven), Agno agent framework
  • AI / ML: Azure OpenAI (private description generation, rewrites)
  • Environment: Windows, virtualenv, .env configuration

Features / Capabilities

  • Markdown as Source of Truth: Contributions stored in MVP-PENDING.md
  • AI-Powered Enrichment: Generates missing private descriptions, rewrites long text (~850 chars)
  • Validation Layer: Normalizes dates, enforces required fields, standardizes tags
  • Browser Automation: Playwright MCP replicates manual form entry
  • Experimental Copilot: Snapshot-driven, prompt-guided workflow with Agno
  • Safe Execution: Requires manual confirmation before submission

Impact & Usage

  • Productivity: Saves hours per reporting cycle
  • Reliability: Prevents errors and enforces consistent formatting
  • Scalability: Extendable for event-specific contributions
  • Exploration: Demonstrates how AI agents can guide browser workflows

Learning Outcomes

  • Integrated Playwright MCP for automation
  • Used Agno for snapshot-driven agent workflows
  • Combined structured parsing with Azure OpenAI enrichment
  • Built validation and retry mechanisms for resilient automation
  • Considered compliance and ethical boundaries
  • Internal Repository: Private
  • Documentation: MVP-PENDING.md template, agent orchestration notes

Important

This project is personal and experimental, created for learning purposes only. It is not endorsed or supported by Microsoft or the MVP Program.

Care must be taken to ensure compliance:

  • Use only with your own data and session.

  • Sharing beyond personal demos may require NDA or compliance review.