Index of every page on this site: llms.txt
All skills
microsoft avatar

/azure-ai

@0d60658
by microsoftmicrosoft/skills3.2k stars
373
Run it yourself

Build and troubleshoot Azure AI Search, Speech, Document Intelligence, Content Understanding, Translator, Content Safety, and OpenAI apps. WHEN: AI Search, vector search, hybrid search, semantic search, indexer, speech-to-text, text-to-speech, fast or batch transcription, Speech SDK logs, session ID, result ID, OCR, document translation, glossary.

Use this Skill: https://skilld.dev/gh/microsoft/skills/azure-ai

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

referencesfoundry-toolsprivate-network.md

≈1.5k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Private networking for Document Intelligence and Content Understanding

For 403 network denials, timeouts, public DNS answers despite a private endpoint, Studio pages that load but cannot call the resource, and service-to-storage fetch failures. Building the network belongs to microsoft-foundry.

First decide the direction

Direction Typical symptom Owner of the fix
Your client/app → DI or CU endpoint 403, timeout, public IP from nslookup Resource network rules, private endpoint, DNS
Browser → Studio / Foundry portal → resource and Blob Page loads, calls fail; status 0 or 403 in the browser Network tab Network path of the browser machine; Blob CORS and firewall (Studio and storage)
DI or CU service → your Blob URL ContentSourceNotAccessible, training cannot read data Storage firewall, managed identity role, SAS; or send bytes instead of a URL

Facts to check against

  • One private endpoint (subresource account) serves every host name of the resource. Create and link a private DNS zone for each host your workload uses: privatelink.cognitiveservices.azure.com, privatelink.openai.azure.com, privatelink.services.ai.azure.com. Learn samples call DI and CU through both <name>.cognitiveservices.azure.com and <name>.services.ai.azure.com, so a missing zone breaks only the host that depends on it.
  • Private endpoint calls need the custom subdomain endpoint; custom DNS servers must forward the privatelink zones to 168.63.129.16.
  • publicNetworkAccess: Disabled accepts private endpoint traffic only. Enabled with networkAcls.defaultAction: Deny allows only listed IP and virtual network rules; enabling rules also blocks Azure services and portal calls that are not listed.
  • IP rules: public IPv4 only, up to 100, no /31 or /32 prefixes (add single addresses), no private ranges. The address that counts is the public egress (NAT, proxy, firewall), not the client's private IP.
  • Virtual network rules need the custom subdomain and the Microsoft.CognitiveServices service endpoint on the subnet; traffic from that subnet then arrives with a private source, so existing IP rules stop matching it.
  • Learn documents the trusted-services bypass for Azure OpenAI only; do not promise it for DI or CU calls.
  • Data loss prevention (restrictOutboundNetworkAccess: true plus allowedFqdnList, up to 15 minutes to apply) limits which hosts the service may fetch URL sources from; Learn lists DI, not CU.
  • Studio from a private network: DI Studio needs the browser to reach the DI resource and the storage account (or the documented Studio IP for auto label). For CU Studio and the Foundry portal allow outbound service tags AzureActiveDirectory, AzureFrontDoor.Frontend, AzureResourceManager, CognitiveServicesManagement and CognitiveServicesFrontEnd (Learn lists them for the Foundry portal only). With public access disabled, use VPN, ExpressRoute or Bastion.

Workflow

  1. Export the account (Azure MCP has no network tools; foundryextensions_resource_get lacks network fields): az cognitiveservices account show -n <name> -g <rg> -o json > account.json.

  2. On the affected client (not your workstation, unless that is the client), run the network checker (Bash / PowerShell; without Node see manual checks):

    node "<skill-directory>/scripts/check-foundry-tools-network.mjs" --account account.json --resolve --connect --client-ip <public-egress-ip>
    & node "<skill-directory>\scripts\check-foundry-tools-network.mjs" --account account.json --resolve --connect --client-ip <public-egress-ip>

    Launcher exit codes are 0 report/help, 1 file/read/JSON error, 2 usage error and 3 Node.js 22+ unavailable. The checker reads UTF-8, UTF-8-with-BOM and UTF-16 PowerShell 5.1 exports.

    --resolve labels each host private, public or not_found; --connect stops at a TLS handshake (no HTTP, no credentials). No IPs or host names are printed. --client-ip only compares a public egress rule; it does not prove which route the failing request used. With public access disabled, trust the private DNS and TLS results for the affected client.

  3. Read-only follow-ups: az network private-endpoint-connection list --id <resource-id>, az network private-dns zone list, az network private-dns link vnet list -g <rg> -z <zone>, az network private-endpoint dns-zone-group list --endpoint-name <pe> -g <rg>.

  4. Propose rule, DNS or endpoint changes as a plan; apply only after the user confirms. Do not open storage or the resource to the internet as a test.

Evidence boundary

Configured state does not prove the path a request took. A host that resolves privately on one machine can resolve publicly from the app's subnet, a proxy or a browser with secure DNS. allowed RBAC plus 403 points to the network; 401 points to authentication (access and roles).

Sources (retrieved 2026-11-18): Private endpoint DNS values · Configure virtual networks for Foundry Tools · Foundry private link · CU security · DI secured access · Data loss prevention · VNet service endpoints

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 warning11d4 checks · Risk SAFE
  • Gen Agent Trust Hub11d

    This skill provides tools and guidance for developing and troubleshooting Azure AI services. It includes a local diagnostic script for analyzing Speech SDK logs with robust safety controls, along with comprehensive references for official Azure SDKs and secure authentication practices.

  • Socket11d

    No alerts

  • Snyk11d

    Risk: LOW · No issues

  • Runlayer9mo

    15/15 files flagged

Signed by skilld at 0d60658. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 2 days ago
metadata
{
  "author": "Microsoft",
  "version": "1.2.6"
}

README badge

README badge for microsoft/skills/azure-ai