More screenshots
Works with the AI Agents you already use
Azure AI: Foundry, Azure OpenAI & RAG in Production
- **Configures secure authentication** using Entra ID and managed identities to replace vulnerable API keys.
Free
Azure AI: Foundry, Azure OpenAI & RAG in Production
Example session with this skill installed
Our app uses OpenAI directly. Move it to Azure in Sweden Central with no API keys in code, and tell me why I keep getting 429s in the load test.
- Read your context and instructions
- Compiled the azure ai
Migration plan: OpenAI -> Azure (Sweden Central)
Summary: one Foundry resource, one chat deployment (Data Zone EU for residency), keyless auth via managed identity, retry + quota fix for the 429s. Code change is 3 lines.
1. Provision
az group create -n rg-ai -l swedencentral
az cognitiveservices account create -n acme-ai -g rg-ai -l swedencentral --kind AIServices --sku S0 --custom-domain acme-ai
az cognitiveservices account deployment create -n acme-ai -g rg-ai --deployment-name chat \
--model-name <model> --model-version <version> --model-format OpenAI --sku-name DataZoneStandard --sku-capacity 100
az role assignment create --assignee <app-managed-identity-id> --role "Cognitive Services OpenAI User" \
--scope $(az cognitiveservices account show -n acme-ai -g rg-ai --query id -o tsv)
2. Code change (keyless)
from openai import OpenAI
from azure.identity import DefaultAzureCredential, get_bearer_token_provider
token = get_bearer_token_provider(DefaultAzureCredential(), "https://cognitiveservices.azure.com/.default")
client = OpenAI(base_url="https://acme-ai.openai.azure.com/openai/v1/", api_key=token)
# model= now the DEPLOYMENT name: "chat"
3. Why the 429s
| Finding | Evidence | Fix |
|---|---|---|
| Quota too low | Load test peaks ~140k TPM vs 50k deployed | Raise to 150k TPM or add a 2nd deployment |
| No backoff | Client retries instantly, 0 waits | max_retries=5, honour retry-after |
| Huge max_tokens | 4,000 reserved per call, avg use 380 | Set max_output_tokens=800 |
4. Verify
python azure_ai_check.py --endpoint https://acme-ai.openai.azure.com --deployment chat
OK auth=Entra ID deployment=chat latency=412 ms tokens in/out=9/2
Next steps: private endpoint + disable public access, diagnostic logs to Log Analytics, budget alert at 80%.
Connects securely to your tools. The creator never sees your data.
About this skill
Enterprise AI deployment on Azure often fails at the networking and identity layer. This skill provides the specific patterns needed to move from basic OpenAI calls to production-ready Azure AI Foundry architectures.
What it does
- Architects RAG pipelines using Azure AI Search with hybrid search and semantic ranking configurations.
- Generates provisioning scripts via az CLI for AIServices resources, model deployments, and role assignments.
- Configures secure authentication using Entra ID and managed identities to replace vulnerable API keys.
- Implements production safeguards including 429 handling, exponential backoff, and content filter management.
- Optimizes deployment strategy by selecting between Global Standard, Data Zone residency, or Provisioned Throughput (PTU).
How it works
- Define the requirement such as a RAG system, a specific model deployment (Llama, GPT, DeepSeek), or a multi-region failover.
- Select the deployment type based on your residency needs and expected TPM/RPM volume.
- Execute provisioning commands to spin up resources and assign the "Cognitive Services OpenAI User" role to your identity.
- Run smoke tests using the provided
azure_ai_check.pyscript to verify endpoint connectivity and latency.
Frameworks & tools
Azure AI Foundry, Azure OpenAI, Azure AI Search, Document Intelligence, Azure CLI, Bicep, Python OpenAI SDK, and Azure Identity library.
Why this beats prompting it yourself
Generic AI advice misses the "Azure way" of handling networking and identity. This skill enforces keyless authentication and correct base_url formatting that standard prompts often hallucinate, saving hours of debugging 401 and 404 errors.
Use cases
- Migrating an existing OpenAI application to Azure enterprise infrastructure.
- Building a RAG system that respects document-level security and ACLs.
- Provisioning model deployments across multiple regions to bypass TPM quotas.
- Configuring private endpoints for regulated data environments.
Known limitations
Does not support legacy Azure OpenAI resource types. Users must manually verify regional model availability as Microsoft quotas and SKU availability change frequently.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for LLM Ops.
Requires existing Azure subscription and CLI access. Model availability varies by region. Does not automate Bicep/Terraform state management.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 6 days ago
- Free with an Agensi account
- Accounts
Azure
Permissions
Allowed Hosts