DEV Community: Sanil Shet

AI Coding: A Powerful Tool - With Risks You Can’t Ignore

Sanil Shet — Thu, 18 Dec 2025 04:54:49 +0000

AI-assisted coding can supercharge development—speeding up tasks, reducing repetitive work, and unblocking progress. But here’s the catch: the moment you stop reviewing every step, it can create more technical debt than progress.

Example:
I recently asked an AI tool to change a database column from integer to varchar without touching existing data. Instead, it dropped the original column, created a new one, and filled it with UUIDs—exactly what I said not to do.
Luckily, this was a development database, so the damage was reversible.

Lesson Learned:
AI is a powerful accelerator, but without guardrails, it can push your project off a cliff just as quickly as it can move it forward.

Why Review Is Critical:
• AI doesn’t fully understand business context or hidden dependencies.
• Schema changes, migrations, and integrations often have cascading effects.
• A single unchecked step can lead to data loss, broken links, or costly rework.

Always validate AI-generated changes—especially in production-critical areas. Think of AI as a junior developer: fast and helpful, but needs supervision.

Design for Business Requirements

Sanil Shet — Thu, 18 Sep 2025 13:34:24 +0000

There is no perfect architecture for any application, Architecture must be designed based on the functional and non-functional requirements , here are few mentions before thinking to design architecture for your application

Every design decision must be justified by the business requirements.
Engineers are tend to be over-engineer, by approaching latest architecture styles and patterns .
Application tends to be eventually experimental by incorporating all the latest fancy tools and design patterns.
Drive your architecture by business needs.
Focus on constraints, assumptions and limits of your application and clearly define business requirements .
Start and Grow application with Metrics in mind

How many concurrent users application can handle?

What is the minimum latency expected?

What is the acceptable outage application can have?

Follow the below basic workflow before building or thinking any architecture for an application , Problem -> Learn -> Design -> Adapt -> Evaluate.
Explanation of the Workflow
Problem: Identify the problem that needs to be solved.
Learn: Gather requirements and understand constraints.
Design: Create an initial design and review it with stakeholders.
Adapt: Make necessary adjustments and implement changes.
Evaluate: Test the solution, collect feedback, and iterate if necessary.

Be a Master in Debugging

Sanil Shet — Mon, 08 Sep 2025 12:27:18 +0000

Ways to improve debugging skills

Every developer needs to find and fix the bug. But how can you squash a defect that you don't know how to find it.
From my personal experience in Build/RUN projects , and in debugging and diagnosis the problem , here are the some ways you can improve your debugging skills .

Approach
You need to go back or in history where it worked correctly. Approach should be go to previous version of the code where it worked and add code segments until it doesn't work anymore . Debugging approach should be more of scientific method . Create a hypothesis , create a experiment to disapprove the hypothesis , repeat it until you find it. From my personal experience in debugging a defect this has helped me a lot to find the root cause and provide solutions to the customer in faster way.

Learn & Adapt
Learn from the bug. One should not repeat the same mistakes. Take the time to learn lessons. Whenever you go through a crisis kind of issues or bugs , some meta skills can help you to improve your debugging skills like "How could I have prevented that? What are the programming techniques that would have helped me to prevent it at first time ?

Follow this and you will be master in doing diagnosis of a problem

[Boost]

Sanil Shet — Wed, 09 Jul 2025 18:29:15 +0000

Sanil Shet

Jul 9 '25

🚀 White Paper: Automate Jira & Confluence with AI: Open Source MCP Server

#mcp #genai

Comments 3

3 min read

🚀 White Paper: Automate Jira & Confluence with AI: Open Source MCP Server

Sanil Shet — Wed, 09 Jul 2025 05:11:58 +0000

White Paper: MCP Server for AI-Powered Jira & Confluence Automation

Author: Sanil Sheth

Table of Contents

Purpose of this Paper
Table of Contents
1. Purpose of this Paper
2. Executive Summary
3. Problem Statement
4. Technical Architecture
5. Key Features & Innovations
6. Implementation Details
7. Business Impact & Value Proposition
8. Risks & Limitations
9. Future Roadmap
10. Conclusion & Key Takeaways
Purpose of this Paper

This white paper aims to inform technology leaders, project managers, and enterprise architects about the design, benefits, and business impact of the MCP server for AI-powered Jira and Confluence automation. It provides a comprehensive overview of the technical solution, its strategic value, and actionable insights for organizations seeking to modernize their project management workflows.
Executive Summary

This white paper presents an innovative solution leveraging a Model Context Protocol (MCP) server to automate and enhance project management workflows in enterprise environments. By integrating Jira, Confluence, and Agentic AI, the MCP server streamlines ticket creation, documentation, and developer assistance, driving productivity and reducing manual effort.
Problem Statement

Modern enterprises face challenges in managing project information, tracking issues, and maintaining documentation across disparate systems. Manual processes are error-prone, time-consuming, and hinder collaboration. There is a need for an intelligent, secure, and extensible platform that automates these workflows and provides actionable insights.
Technical Architecture
5.1 MCP Server
- Node.js/TypeScript backend
- Registers tools for Jira and Confluence integration
- Exposes RESTful endpoints for ticket creation, updates, and queries
5.2 Integrations
- Jira: Fetches, creates, and updates issues; supports real-time status and prioritization
- Confluence: Updates pages by partial title match; automates meeting note documentation
- Agentic AI (GitHub AI): Provides reasoning, summarization, and automation via prompt engineering
5.3 Security
- Sensitive information redaction filter (emails, secrets, tokens, etc.) before sending data to LLMs
- Configurable proxy and credential management
Key Features & Innovations
- Automated Meeting-to-Jira Ticket Conversion: Instantly turns meeting notes into actionable Jira tickets using AI
- Smart Documentation: Updates Confluence pages with context-aware content
- AI-Powered Suggestions: Summarizes issues, recommends priorities, and generates comments
- Real-Time Status & Highlighting: Displays new and urgent issues for rapid response
- Security Filter: Ensures privacy by redacting sensitive information before AI processing
Implementation Details
7.1 Workflow
1. User submits meeting notes via chatbot or VS Code extension
2. MCP server redacts sensitive info and extracts keywords
3. Related Jira issues are fetched and summarized
4. Agentic AI generates actionable items and ticket details
5. MCP server creates Jira tickets and updates Confluence documentation
7.2 Extensibility
- Modular tool registration for new integrations
- Configurable AI models and endpoints
- Pluggable security filters
Business Impact & Value Proposition
8.1 Productivity Gains
- Automated ticket creation reduces manual entry time by up to 80%
- Smart documentation minimizes information loss and context switching
- AI-powered insights help teams prioritize and resolve issues faster
8.2 Cost Savings
- Reduced operational overhead and labor costs
- Faster onboarding with AI-generated summaries
- Minimized rework through accurate, real-time documentation
8.3 Competitive Advantage
- Agility and rapid response to project changes
- Innovation enablement by freeing developer time
- Data-driven decisions with AI analytics
8.4 Risk Mitigation
- Security and compliance via automated redaction
- Auditability and traceability of all actions
- Reduced vendor lock-in with modular architecture
8.5 Stakeholder Benefits
- Project managers: Real-time dashboards and reduced admin burden
- Developers: Context-rich tickets and instant documentation
- Executives: Improved visibility and measurable ROI
8.6 Strategic Alignment
- Supports digital transformation and process optimization
- Scalable for growing teams and complexity
- Sustainable for remote/hybrid work models
Risks & Limitations
- AI quota limits and service availability
- Reliance on third-party APIs (Jira, Confluence, GitHub AI)
- Need for ongoing security updates and filter enhancements
Future Roadmap
- Support for additional project management tools
- Advanced analytics and reporting
- Customizable AI workflows and prompt templates
- Enhanced security and compliance features
Conclusion & Key Takeaways

The MCP server represents a significant step forward in automating and securing enterprise project management workflows. By combining robust integrations, AI-powered automation, and privacy-first design, it delivers measurable productivity gains and sets the foundation for future innovation.

Key Takeaways:

MCP server enables seamless, secure automation for Jira and Confluence, powered by AI.
Organizations benefit from faster ticketing, smarter documentation, and reduced manual effort.
Built-in security and extensibility make it suitable for enterprise adoption and future growth.
Adopting MCP server accelerates digital transformation and positions teams for ongoing innovation.