← All runbooks
microsoft / microsoft-foundry

Microsoft Foundry Skill

This runbook operationali

agent claude-codemodel claude-sonnet-4-6snapshot python312-uveval programmatic7 stepsv1.0.0

Deploy Microsoft Foundry Skill to your jetty.io

One-click installs this runbook into a collection on your Jetty account. You can run it from the Spot dashboard, schedule it, or pipe inputs in via the API.

The shape of the run

7 steps · start to finish.

  1. 1
    Step 1

    Environment Setup

    Verify the required tools are available and the Azure session is authenticated.

    echo "=== Foundry Skill — Environment Check ==="
    command -v az   >/dev/null 2>&1 && echo "PASS: az CLI found"   || echo "WARN: az CLI not found — install Azure CLI"
    command -v azd  >/dev/null 2>&1 && echo "PASS: azd found"      || echo "INFO: azd not found — optional for bootstrapping"
    command -v docker >/dev/null 2>&1 && echo "PASS: docker found" || echo "INFO: docker not found — required for hosted agent deploy"
    
    # Verify Azure login
    az account show >/dev/null 2>&1 && echo "PASS: Azure session active" || { echo "FAIL: Not logged in to Azure"; exit 1; }
    az account show --query "{subscription:id, name:name}" -o json
    
    mkdir -p /app/results
    echo "Results dir ready: /app/results"
    

  2. 2
    Step 2

    Discover Foundry MCP Entry Point

    **MANDATORY**: Before executing any workflow-specific steps, call the Azure MCP `foundry` tool to discover available Foundry MCP tools and their parameters. Treat this as a discovery/help step.

  3. 3
    Step 3

    Load Sub-Skill Document

    **MANDATORY**: Before executing workflow-specific steps, read the corresponding sub-skill document. Do not call workflow-specific MCP tools without first reading its skill document.

  4. 4
    Step 4

    Resolve Agent Context

    Run context resolution only for values not already provided by the user.

  5. 5
    Step 5

    Execute Sub-Skill Workflow

    Execute the workflow defined by the sub-skill loaded in Step 3. Follow the sub-skill document exactly — do not skip pre-checks or validation logic.

  6. 6
    Step 6

    Iterate on Errors (max 3 rounds)

    If any workflow step fails:

  7. 7
    Step 7

    Write Executive Summary

    Write `/app/results/summary.md` with the run outcome, sub-skill used, agent context, and any issues.