azure-ai

Use for Azure AI: Search, Speech, OpenAI, Document Intelligence. Helps with search, vector/hybrid search, speech-to-text, text-to-speech, transcription, OCR. WHEN: AI Search, query search, vector search, hybrid search, semantic search, speech-to-text, text-to-speech, transcribe, OCR, convert text to speech.

Install with Tessl CLI

npx tessl i github:microsoft/azure-skills --skill azure-ai

What are skills?

2.02x

Quality

77%

Does it follow best practices?

Impact

99%

2.02x

Average score across 3 eval scenarios

Optimize this skill with Tessl

npx tessl skill review --optimize ./.github/plugins/azure-skills/skills/azure-ai/SKILL.md

SKILL.md

Review

Evals

Evaluation results

100%

48%

Product Catalog Search Service

Azure AI Search Python: vector index setup, batch upload, hybrid search, auth pattern

Criteria

Without context

With context

Production credential

100%

Dev credential

100%

Credential env check

100%

No hardcoded secrets

100%

BufferedSender for batch

100%

Vector field type

100%

Vector field properties

100%

HNSW algorithm

100%

Hybrid search query

100%

Semantic ranking

100%

Idempotent index creation

100%

Without context: $0.5399 · 2m 20s · 27 turns · 34 in / 8,790 out tokens

With context: $0.8256 · 2m 48s · 36 turns · 43 in / 10,376 out tokens

98%

28%

Legal Document Summarization Service

Azure OpenAI .NET: client hierarchy, streaming, rate limit handling, auth

Criteria

Without context

With context

AzureOpenAIClient root

100%

GetChatClient hierarchy

100%

Client instance reuse

100%

75%

Streaming for long output

100%

Exponential backoff on 429

100%

o1: DeveloperChatMessage

100%

o1: ReasoningEffortLevel

100%

Entra ID auth

100%

Production vs dev credential

100%

Config from env vars

100%

Without context: $0.3144 · 1m 36s · 14 turns · 16 in / 6,096 out tokens

With context: $0.5013 · 1m 44s · 25 turns · 63 in / 6,380 out tokens

100%

74%

Automated Invoice Data Extraction

Azure Document Intelligence TypeScript: REST client patterns, async polling, error handling

Criteria

Without context

With context

Client as function

100%

getLongRunningPoller

100%

pollUntilDone

100%

base64Source for local files

100%

isUnexpected check

100%

paginate() for listing

100%

Correct package imports

100%

Correct analyze path

100%

Environment config

100%

prebuilt-invoice model

100%

Without context: $0.4048 · 1m 37s · 21 turns · 67 in / 6,247 out tokens

With context: $0.6543 · 2m 7s · 30 turns · 34 in / 8,074 out tokens

Evaluated: 4 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Product Catalog Search Service Legal Document Summarization Service Automated Invoice Data Extraction

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.