How-To Guides

Choose a task, run it with Copilot, and check the required output independently. These guides also cover the controls and evidence needed to investigate a failure without treating the assistant's claims as verification.

Guides

• Complete Example — Run a self-contained banking-tool test
• Test MCP Servers — Attach tools and verify operation data
• Test CLI Tools — Run a Git task and verify repository state
• Inspect Execution Evidence — Let your coding agent investigate pytest failures and saved JSON
• Cost Estimation — Configure rates and distinguish estimates from billing
• Test Plugins — Load a configuration bundle and verify its output
• Test Skills and Custom Agents — Configure bounded tasks and guarded application tools
• Tool-returned images — Check image capture and extract saved bytes
• CI/CD integration — Authorize live jobs and preserve pytest and JSON evidence