$ loading_
Deploy, evaluate, fine-tune, and optimize Microsoft Foundry agents and models.
The material indicates this skill is primarily prompt/workflow documentation; system signals also mark it as prompt-only, open-source, and sourced from a Microsoft GitHub repository, so overall risk is low. It may guide later use of the Azure MCP foundry tool for deployment and resource management, but this material itself does not declare built-in secrets, remote endpoints, or local execution capability.
The material and system checks show that this skill itself does not require secrets or environment variables, and there is no stated behavior to collect, store, or transmit credentials. Note that Azure/Foundry workflows may involve cloud credentials when external Azure MCP tools are actually used, but that is not a direct requirement of this skill material.
No remote endpoints are declared in the current material, and the system marks host as none; as a prompt-only skill, it does not itself show active networking or data exfiltration behavior. The docs mention later use of the Azure MCP foundry tool, but this material does not specify any additional unknown egress targets.
This item is marked prompt-only, and the README mainly contains workflow rules and sub-skill routing; there is no indication that the skill itself starts local processes, executes scripts, or invokes system capabilities. Docker build, ACR push, and deployment are described as external workflow capabilities it may guide, not actions directly executed by this skill material itself.
The material does not state that this skill itself can read or write local files, databases, or other resources, nor does it request broad data permissions. Although it mentions traces, dataset curation, logs, and telemetry, those appear to be data types handled by downstream external tools rather than direct data access implemented by this skill text.
The source is an open-source Microsoft GitHub repository, which is auditable, and it has some community adoption (222 stars); these are clear risk-reducing signals. License and maintenance activity are not clearly stated in the material, which is a minor information gap, but not enough to outweigh the positive evidence from the official open-source source.
Copy the install command and let the AI configure it · recommended for beginners
Please install the "microsoft-foundry" skill from askskill: 1. Download https://raw.githubusercontent.com/microsoft/GitHub-Copilot-for-Azure/main/plugin/skills/microsoft-foundry/SKILL.md 2. Save it as ~/.claude/skills/microsoft-foundry/SKILL.md 3. Reload skills and tell me it's ready
Guide me through deploying an agent in Microsoft Foundry: from Docker build and ACR push to creating the Foundry project/resource and a hosted agent, including required RBAC permissions, region, and capacity checks.
A step-by-step deployment guide with prerequisites, sample commands, permission setup, and troubleshooting advice.
I already have a Foundry agent. Help me design a batch and continuous evaluation plan, including dataset preparation, grader setup, key metrics, continuous monitoring workflow, and how to improve prompts and instructions from the results.
A complete evaluation plan covering datasets, metric definitions, automation ideas, and optimization recommendations.
Help me plan model fine-tuning and agent optimization in Microsoft Foundry: compare when to use SFT, DPO, and RFT, explain training data preparation, how to curate datasets from traces, and how to use the prompt optimizer and Agent Optimizer scaffold to improve results.
A fine-tuning and optimization roadmap with method selection guidance, data preparation steps, and an actionable improvement workflow.
This skill helps developers work with Microsoft Foundry resources, covering model discovery and deployment, complete dev lifecycle of AI agent, evaluation workflows, and troubleshooting.
MANDATORY: Before executing ANY workflow, you MUST first call the Azure MCP
foundrytool and inspect the available Foundry MCP tools and related parameters. Treat this initialfoundrycall as a discovery/help step. For this skill, Azure MCPfoundryis the required entry point for Foundry-related MCP operations.
MANDATORY: Before executing ANY workflow-specific steps, you MUST read the corresponding sub-skill document. Do not call workflow-specific MCP tools for a workflow without reading its skill document. This applies even if you already know the MCP tool parameters — the skill document contains required workflow steps, pre-checks, and validation logic that must be followed. This rule applies on every new user message that triggers a different workflow, even if the skill is already loaded.
This skill includes specialized sub-skills for specific workflows. Use these instead of the main skill when they match your task:
| Sub-Skill | When to Use | Reference |
|---|---|---|
| deploy | Containerize, build, push to ACR, create/update/clone agent deployments | deploy |
| invoke | Send messages to an agent, single or multi-turn conversations | invoke |
| invocations-ws | Build, deploy, and connect to hosted agents that speak the invocations_ws duplex WebSocket protocol — voice agents, real-time streams, and signaling for out-of-band media transports. | invocations-ws |
| observe | Evaluate agent quality, run batch evals, analyze failures, optimize prompts, improve agent instructions, compare versions, set up CI/CD monitoring, and enable continuous production evaluation | observe |
| trace | Query traces, analyze latency/failures, correlate eval results to specific responses via App Insights customEvents | trace |
| troubleshoot | View hosted agent logs, query telemetry, diagnose failures | troubleshoot |
| create | Create new hosted agent applications. Supports Microsoft Agent Framework, LangGraph, or custom frameworks in Python or C#, across responses, invocations, or invocations_ws protocols. | create |
| agent-optimizer | Make existing Python hosted-agent code optimization-ready, configure eval.yaml, run Agent Optimizer jobs, apply candidates locally, and deploy through azd after review. | agent-optimizer |
| eval-datasets | Harvest production traces into evaluation datasets, manage dataset versions and splits, track evaluation metrics over time, detect regressions, and maintain full lineage from trace to deployment. Use for: create dataset from traces, dataset versioning, evaluation trending, regression detection, dataset comparison, eval lineage. | eval-datasets |
| project/create | Creating a new Azure AI Foundry project for hosting agents and models. Use when onboarding to Foundry or setting up new infrastructure. | project/create/create-foundry-project.md |
| resource/create | Creating Azure AI Services multi-service resource (Foundry resource) using Azure CLI. Use when manually provisioning AI Services resources with granular control. | resource/create/create-foundry-resource.md |
…
Set up Entra ID app registration, OAuth, permissions, and MSAL integration.
Set up Azure Application Insights telemetry for web apps with best practices.
Troubleshoot Azure Event Hubs and Service Bus SDK connection and messaging issues.
Find, list, and inspect Azure resources across subscriptions and resource groups.
Deploy prepared Azure apps with built-in recovery for common release errors.
Use Azure AI for search, speech, transcription, and OCR tasks.
Fine-tune, deploy, and evaluate models in Azure AI Foundry.
Deploy Azure OpenAI models and find regional capacity intelligently.
Explore AI models, MCP servers, and agent integrations in one gateway.
Expose Azure ML endpoints as MCP tools for natural-language AI agent calls.
Guide custom Azure OpenAI deployments with detailed model and capacity settings.
Give AI real-time Azure infrastructure access for ops checks, policy validation, and Terraform analysis.