Project: Creative Apps - Agile Agents #121
Description
Track
Creative Apps (GitHub Copilot)
Project Name
Agile Agents
GitHub Username
Repository URL
https://github.com/SamuelCano03/agileagents
Project Description
Agile Agents is a virtual agile squad integrated into VS Code via the Model Context Protocol (MCP). It coordinates specialized agents, such as a Scrum Master and Product Owner, to manage software development lifecycles directly within GitHub Copilot Chat. The system leverages the Jira API to retrieve real-time information and context directly from the Jira board, including active sprint issues and detailed ticket metadata. This integration allows developers to manage their backlog and update statuses without leaving their IDE, significantly reducing operational friction.
To mitigate execution risks, the project implements a mandatory "Plan & Apply" governance model for all write operations, ensuring no accidental changes are made to the Jira board without explicit human confirmation. Furthermore, it ensures that impediments identified during dailies are immediately registered and followed up through automated blocker detection and session persistence. Key features include a native multi-turn daily flow that requests updates within a structured conversation. Finally, the system offers full Jira integration for transitioning and editing issues, supported by traceable auditing through dedicated logs for every planning and execution event.
Demo Video or Screenshots
Canva presentation (In progress): https://www.canva.com/design/DAHCwzNW03I/kZLQrn5uLOHZbtuM6-HgUQ/edit?utm_content=DAHCwzNW03I&utm_campaign=designshare&utm_medium=link2&utm_source=sharebutton
Planning with Copilot Plan Mode:]
Creating accounts for Alice and Bob:3
Use of Copilot:
Test:
Primary Programming Language
Python
Key Technologies Used
- GitHub Copilot & VS Code: Main interface and development support.
- Model Context Protocol (MCP): Native bridge for connecting AI with local tools and third-party APIs.
- Jira Cloud API: Full integration for software lifecycle management.
Submission Type
Individual
Team Members
- @SamuelCano03 - Role (Everything)
Submission Requirements
- My project meets the track-specific challenge requirements
- My repository includes a comprehensive README.md with setup instructions
- My code does not contain hardcoded API keys or secrets
- I have included demo materials (video or screenshots)
- My project is my own work with proper attribution for any third-party code
- I agree to the Code of Conduct
- I have read and agree to the Disclaimer
- My submission does NOT contain any confidential, proprietary, or sensitive information
- I confirm I have the rights to submit this content and grant the necessary licenses
Quick Setup Summary
- Clone the repo:
https://github.com/SamuelCano03/agileagents - Install dependencies (uv):
uv sync - Configure environment variables: copy .env.sample to .env and fill required values
- Run the project (MCP server):
uv run uvicorn jira_mcp_server.server:app --port 8000 - Check the agileAgentsJira on the Copilot tools
- Open a Copilot VS Chat and ask it to start a daily or request specific tasks, e.g.:
update <ISSUE-KEY> status to 'Done', or - Optional (CLI mode):
uv run -m src.interfaces.cli --sm-assistant-chat --language en
Technical Highlights
I am most proud of how this implementation optimizes the developer's workflow by eliminating context switching. By allowing team members to update Jira issues directly from their IDE, we remove the friction of jumping between tools for minor updates. Additionally, the feature that simulates a “daily stand-up” is a key technical highlight; it helps developers identify dependencies and prepare their updates ahead of time, ensuring more productive team syncs.
I’m also particularly proud of how we combined traditional REST APIs with Jira’s MCP to build a more flexible and powerful integration layer. This hybrid approach significantly improved the robustness of the project, made the interactions with Jira more expressive, and unlocked more natural workflows for developers interacting with project management data.
Finally, having a personal Scrum Master Assistant available at any time allows developers to quickly clarify Scrum concepts and best practices, reducing friction and supporting continuous learning within the workflow.
Challenges & Learnings
One of the main challenges was deploying the model. I initially tried to set it up using Azure AI Foundry, but the setup and configuration overhead made the process slower and more complex than expected. In the end, switching to GitHub Models turned out to be a much simpler and more cost-effective solution, which helped me iterate faster and focus on the product instead of the infrastructure.
Another challenge was designing a consistent and reliable flow for the daily standup interactions. Making sure the assistant followed a clear structure while still being flexible for different types of requests required several iterations and refinements.
Security and action safety were also important concerns. Implementing a two-phase contract with plan_action and apply_action helped ensure that actions were visible, reviewable, and intentional before being executed. This approach improved both transparency and safety, and gave me more confidence in allowing the assistant to interact with external systems.
Overall, these challenges taught me the importance of choosing simple tooling early on, iterating on conversational flows with real usage in mind, and designing explicit safety mechanisms when building agents that can take actions.
Contact Information
saelcc03@gmail.com / U202116508@upc.edu.pe
Country/Region
Peru