Skip to content

COPILOT.md

Latest commit

 

History

History
103 lines (69 loc) · 4.22 KB

COPILOT.md

File metadata and controls

103 lines (69 loc) · 4.22 KB
Raw

Development with GitHub Copilot

This document describes how GitHub Copilot was used to build AgriValue Tracker for Battle #1: Creative Apps in the AI Agents League.

How Copilot Accelerated Development

1. Backend scaffolding (server.js)

Copilot Chat was used to scaffold the Express API structure:

Create an Express API with routes for crop market data, quality evaluation,
and value-chain optimization. Use Azure Key Vault for secrets with env fallback.

Copilot generated the polling loop for Azure AI Foundry agent runs (pollRun), the thread/message/run API sequence, and telemetry logging patterns.

2. Fabric IQ semantic graph (lib/fabric-iq.js)

Copilot inline suggestions helped structure the crop → processor → certification ontology:

  • Entities: crops, processors, compliance rules, organic bonuses
  • Relationships: each processor inputs a crop; each crop has certifications
  • Reasoning: optimizeValueChain() computes Path A vs Path B with quality multipliers

Prompt used in Copilot Chat:

Model a semantic market graph for Sicilian agriculture: 5 crops, local processors,
transport costs, DOP/DOC certifications, and Foundry compliance rules as separate
functions exportable to both Express and MCP.

3. MCP server for Copilot in VS Code (mcp-server.js)

Copilot helped implement the Model Context Protocol server exposing three tools:

Tool Purpose
get_market_intelligence Grounded pricing and compliance for a crop
generate_supply_contract Draft Fair-Trade supply agreements
evaluate_crop_quality Quality grade and profit multiplier

Example Copilot Chat prompt (with MCP enabled):

@agrivalue-iq generate_supply_contract crop="Olive Oil" qtyTons=15 isOrganic=true buyer="Cooperativa Valle del Belice"

Now customize section 4 for EU DOP export requirements and add a force majeure clause.

4. Frontend terminal UI (public/index.html)

Copilot assisted with:

  • Tailwind layout for the "institutional terminal" aesthetic
  • Chart.js market trend visualization wired to Fabric IQ forecast data
  • Visual quality scanner animation and console log feed
  • Contract export and "Copy for Copilot MCP" workflow buttons

5. Debugging with Copilot Chat

When the market chart threw getElementById('trendChart') errors, Copilot Chat identified the missing <canvas> element and suggested wrapping initChart in try/catch for static demo mode.

MCP Setup in VS Code

  1. Install GitHub Copilot and enable MCP in settings
  2. Open this repo in VS Code — .vscode/mcp.json registers the server automatically
  3. Run npm install then verify: npm run mcp (should log "running on stdio")
  4. In Copilot Chat, reference tools with @agrivalue-iq

Suggested Copilot Workflows for Judges

Write a contract from live optimization results

  1. Run the app locally (npm start)
  2. Optimize Olive Oil with organic certification
  3. Click Copy for Copilot MCP in the results panel
  4. Paste into Copilot Chat — MCP generates a grounded contract draft

Extend the semantic graph

@agrivalue-iq get_market_intelligence crop="Grapes"

Add a new processor "Etna Volcanic Wines" in lib/fabric-iq.js with costs
and update the MCP enum in mcp-server.js

What Was Built Manually vs. Copilot-Assisted

Component Copilot role
Azure Foundry agent integration Scaffolding + API version headers
Fabric IQ ontology Data structure + helper functions
MCP tool schemas JSON Schema definitions
UI copy and styling Tailwind classes + animation CSS
Security (Key Vault, .env) Pattern suggestions, manual review
Challenge documentation Structure outline, manual editing

Security Note

Copilot was not given real API keys. All secrets use .env (gitignored) per the challenge security guidelines. See .env.example for required variables.