Copilot Context Keeper

A lightweight context preservation framework for AI-assisted rapid prototyping and vibe-coding sessions. Designed specifically for GitHub Copilot Agent mode (@workspace conversations).

The Copilot Context Keeper framework helps maintain context across GitHub Copilot sessions through 5 lightweight files, enabling quick prototype development without losing momentum between sessions.

Prerequisites

• Have an idea to explore
• Understanding of your collaboration style (git workflow, documentation preferences, etc.) and tech stack
• GitHub Copilot with Agent mode enabled
• Access to the template files in /templates/ folder

File Templates Overview

The Copilot Context Keeper framework uses 5 essential files, each with a specific purpose and ownership:

copilot-context-keeper.md - The AI Contract

👤 User copies | The most important file - this contains the complete framework instructions that tell the AI HOW TO WORK WITH MEMORY BANKING. This file must be copied into every .copilot/vibes/ folder and serves as the AI's operating manual. It defines the file structure, update processes, and memory preservation workflow.

idea.md - Project Concept

🤖 AI creates from user description | Captures the core concept and success criteria for your prototype. This is your north star - created once and rarely changes. The AI creates this from your initial description, but may ask clarifying questions to complete all sections using the template structure.

vibe.md - Collaboration Style

👤 User creates & maintains | Defines how you want to collaborate with the AI assistant. Specifies your interaction style, tech stack preferences, decision-making approach, git workflow, security practices, documentation requirements, and speed vs quality trade-offs.

state.md - Technical Snapshot

🤖 AI creates & maintains | The living technical snapshot of your prototype. Updated frequently by the AI as you build. Contains current stack, architecture overview, file structure, what's working/broken, immediate next steps, and current focus.

decisions.md - Decision Log

🤖 AI creates & maintains | Log of key choices made during development. Prevents re-discussing the same decisions. The AI creates and maintains this file as architectural and technical decisions are made, following the template structure.

How the System Works

This makes the framework complete and self-contained. The GitHub Copilot Agent gets both:

• How to work (from the framework instructions in copilot-context-keeper.md)
• What to work on (from the 4 context files: idea.md, vibe.md, state.md, decisions.md)

Quick Project Setup

Follow these steps to set up a new prototype with Copilot Context Keeper (paths are relative to example project root under the /examples/ folder):

# Create Copilot Context Keeper structure
mkdir -p .copilot/vibes
cd .copilot/vibes

# Copy the AI instructions (tells AI how to use the framework)
cp ../../copilot-context-keeper.md .

# Copy ALL template files (AI uses these as references for consistent formatting)
cp ../../templates/*.md ./

# Customize your collaboration template
# Edit vibe.md to match your preferences

# Return to project root
cd ../..

# Start a new prototype session with GitHub Copilot Agent
# Use @workspace in your GitHub Copilot chat

That's it! The GitHub Copilot Agent will automatically handle:

• Using idea.md template when creating from your prototype description
• Using state.md template when creating and maintaining technical snapshot
• Using decisions.md template when logging decisions
• Following your vibe.md preferences for git workflow, security practices, and collaboration style

Just describe your prototype idea to the Copilot Agent using @workspace, and it will follow the template formats while using your vibe.md preferences.

Real Example: Builder-Tracker

See the complete working example in /examples/builder-tracker/ which demonstrates:

• Starting prompt used to initialize the prototype (using the example prompt below)
• Resume prompt used to continue work across sessions (using the example prompt below)
• Compiled templates showing how the framework files look when filled out
• Complete vibe.md with git workflow, documentation requirements, and collaboration preferences

This example shows the Copilot Context Keeper framework applied to a real prototype project using the exact prompts provided in this README.

Step-by-Step Usage Guide

1. Starting a New Prototype Session

1. Copilot Agent reads the framework instructions (copilot-context-keeper.md) to understand how to use the framework
2. Describe your prototype idea to the Copilot Agent (can be brief - just the core concept)
3. Agent creates idea.md using the template structure:
   • If your description covers all template sections → Agent creates complete idea.md
   • If information is missing → Agent asks clarifying questions to fill gaps
   • You review and approve the final idea.md before proceeding
4. Agent follows your vibe.md preferences for collaboration style, workflow, and project setup
5. Start coding - Agent maintains state.md automatically

See the Example Prompts section below for the exact prompt to use when starting a new project.

2. Collaboration Style Configuration

The Copilot Context Keeper framework is highly configurable through your vibe.md file. You can specify:

• Git workflow preferences (branching, commits, merging)
• Documentation requirements (what docs to create, when)
• Security practices (how to handle secrets, environment variables)
• Decision-making style (what to ask about vs. assume)
• Process management (how to run long-running processes)

The template in /templates/vibe.md includes an example git workflow configuration. Customize it to match your preferred working style.

3. Resuming Work (Key Benefit!)

1. Agent reads the framework instructions (copilot-context-keeper.md) to understand how to work
2. Agent reads all 4 context files (idea.md, vibe.md, state.md, decisions.md) - takes ~30 seconds
3. Agent confirms context: "I see we're building [X], currently working on [Y], next step is [Z]"
4. Agent continues work following your vibe.md preferences
5. No re-explaining needed - context is preserved

See the Example Prompts section below for the exact prompt to use when resuming work.

4. Ending a Session

1. Agent updates state.md with current progress
2. Agent logs any new decisions made during the session
3. Agent follows end-of-session preferences from your vibe.md (e.g., git commits, cleanup)
4. Ready for next session - no context loss

Example Prompts

Starting a New Prototype Session

@workspace Hi! I want to start a new prototype using Copilot Context Keeper. 

Please read the framework instructions in .copilot/vibes/copilot-context-keeper.md first to understand how to work with this system.

My prototype idea: [Describe your idea here - can be brief, just the core concept]

Example: "I want to build a simple web app that helps developers track their daily coding habits. Users can log what they worked on, how long they spent, and rate their productivity. The goal is to identify patterns and improve focus."

Resuming an Existing Project

@workspace Hi! I'm resuming work on my prototype using Copilot Context Keeper.

Please read the framework instructions in .copilot/vibes/copilot-context-keeper.md first, then read all the context files in .copilot/vibes/ folder to understand the current state.

Once you've reviewed everything, please confirm what we're building, where we left off, and what the next steps should be.

Tips for Effective Copilot Agent Sessions

Customizing Your vibe.md

• Be comprehensive - specify every aspect of collaboration (tech stack, git workflow, documentation requirements, etc.)
• Be specific about your preferences to avoid back-and-forth questions
• Define your collaboration style clearly for consistent AI behavior
• Include all working preferences - decision-making style, process management, security practices

Managing File Sizes

• Keep files under the size limits (see main documentation)
• Focus on actionable information only
• Use bullet points instead of long paragraphs
• Remove outdated information regularly

Git Workflow Tips

• Feature branches keep main clean and working
• Regular commits help track progress
• Descriptive commit messages aid in context rebuilding
• Ask before merging maintains code quality

Troubleshooting

"Agent doesn't understand my context"

• Check if all 4 files exist in .copilot/vibes/ folder
• Ensure idea.md clearly describes the core concept
• Update state.md if current status is unclear

"Setup commands don't work"

• Verify template files exist in /templates/ folder
• Check file paths in copy commands
• Ensure you're in the correct directory

"Git workflow is confusing"

• Stick to the feature branch pattern
• Let Agent handle branch creation and management
• Always ask Agent before merging to main

"Files are getting too long"

• Review size limits in main documentation
• Focus on essential information only
• Archive old decisions that are no longer relevant

Advanced Usage

Multiple Prototypes

• Each prototype gets its own .copilot/vibes/ folder
• Reuse the same vibe.md template across projects
• Maintain separate git repositories for each prototype

Collaboration with Others

• Share the entire project folder including .copilot/vibes/
• Others can read the context files to understand the project

Evolution to Production

• Use the prototype as a proof of concept
• Decision log helps avoid repeating the same architectural discussions

Credits and Acknowledgments

This GitHub Copilot Agent Memory Banking framework is inspired by and adapted from the excellent work by Nicola Cremaschini and their original Amazon Q Vibes Memory Banking repository.

🙏 Special Thanks:

• Original Author: Nicola Cremaschini
• Original Repository: amazon-q-vibes-memory-banking
• Inspiration: The brilliant concept of lightweight context preservation for AI-assisted rapid prototyping

The core memory banking methodology, file structure, and workflow patterns are based on Nicola's innovative framework. This adaptation brings the same powerful context preservation capabilities to GitHub Copilot's agent mode.

Differences from Amazon Q Version

This GitHub Copilot Agent version differs from the original Amazon Q version in:

• Uses .copilot/vibes/ instead of .amazonq/vibes/ folder structure
• Optimized prompts for @workspace conversations
• Adapted workflow for GitHub Copilot's agent mode
• Maintains the same core memory banking principles and file structure

Please check out the original repository for the Amazon Q Developer version and to support the original creator's excellent work!

Supporting the Original Work

If you find this framework useful, please also:

• ⭐ Star the original repository: amazon-q-vibes-memory-banking
• 👤 Follow the original author: @ncremaschini
• 🔄 Share both repositories with others who might benefit
• 💡 Consider contributing improvements back to the original project