Skip to content

M1_VALIDATION_GUIDE.md

Latest commit

 

History

History
74 lines (50 loc) · 1.78 KB

M1_VALIDATION_GUIDE.md

File metadata and controls

74 lines (50 loc) · 1.78 KB

M1 Parity Validation Guide

What gets validated

1) MCP server path

The Python repo should mirror the C# lecture flow for Milestone 1:

  1. Start a local stdio MCP server
  2. Complete initialize
  3. Discover get_system_info
  4. Execute get_system_info
  5. Reject an invalid tool name without crashing

2) LLM-backed client path

The Python client should:

  1. Start the same local stdio MCP server
  2. Fetch get_system_info
  3. Build an LLM prompt from that payload
  4. Fail fast with a clear message if Azure OpenAI configuration is missing

Automated checks

Run all tests:

cd /home/alon/MCPPythonDemo
python3 -m unittest discover -s tests -v

Key coverage:

  • tests/test_m1_smoke.py - stdio MCP handshake, tool discovery, happy path, extra args, invalid tool
  • tests/test_system_info.py - graceful fallback when /proc files are missing
  • tests/test_client.py - client prompt building, fake-LLM agent loop, missing-config CLI failure

Smoke checks

Server + agent smoke:

python3 scripts/smoke_test.py

Expected behavior:

  • Server portion prints discovered tools and a live get_system_info payload
  • Agent portion verifies that the client exits with code 2 and emits a human-readable Azure config message when credentials are absent

Required Azure env vars for the agent path

Either prefix style is accepted:

export AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com/"
export AZURE_OPENAI_API_KEY="your-api-key"
export AZURE_OPENAI_DEPLOYMENT="model-router"

or:

export MCP_DEMO_AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com/"
export MCP_DEMO_AZURE_OPENAI_API_KEY="your-api-key"
export MCP_DEMO_AZURE_OPENAI_DEPLOYMENT="model-router"

Optional:

export AZURE_OPENAI_API_VERSION="2024-10-21"