can not get skillset/indexer to work with Microsoft.Skills.Text.SplitSkill and Microsoft.Skills.Text.AzureOpenAIEmbeddingSkill

Hello Jeff Schulz,

Welcome to the Microsoft Q&A and thank you for posting your questions here.

I understand that you can not get skillset/indexer to work with Microsoft.Skills.Text.SplitSkill and Microsoft.Skills.Text.AzureOpenAIEmbeddingSkill.

To resolve the issue:

1. Validate enrichment tree with Debug Sessions before deployment by run Debug Session in Azure portal > capture Inputs and Outputs for targeted skills. - https://learn.microsoft.com/en-us/azure/search/cognitive-search-concept-troubleshooting and https://github.com/MicrosoftDocs/azure-docs/blob/main/articles/search/cognitive-search-concept-troubleshooting.md gives more insights. Then, confirm actual node paths used (e.g., /document/pages/0/text, /document/Summary).
2. Adjust JSON schema strictly:
   • Using Base template: Skillset JSON schema docs
   • Provide full example for SplitSkill + embedding:

      "name": "splitPages",
      "context": "/document",
      "inputs": [...],
      "outputs": [
       { "name": "textItems", "targetName": "pages" },
       { "name": "ordinalPositions", "targetName": "pageOrdinal" }
      ],
      // -------
      {
      "name": "embedPages",
      "context": "/document/pages/*",
      "inputs": [{ "name": "text", "source": "/document/pages/*/text" }],
      "outputs": [{ "name": "embedding", "targetName": "pageEmbedding" }]
      }
   

3. Correct outputFieldMappings for an example:

      "outputFieldMappings": [
        { "sourceFieldName": "/document/pages/*/text",
          "targetFieldName": "SummaryChunk" },
        { "sourceFieldName": "/document/pages/*/pageEmbedding",
          "targetFieldName": "VectorEmbedding" }
      ]
   

   Use concat(TitleID, '_', pageOrdinal) for ChunkID.
4. Add ConditionalSkill to skip records:

      {
        "@odata.type": "#Microsoft.Skills.Text.ConditionalSkill",
        "conditions": [
          {
            "if": { "source": "/document/Summary", "operator": "Exists" }
          }
        ],
        "trueCase": { "skills": [ /* call next skills only if exists */ ] }
      }
   

5. Use Debug Session to verify null filtering
   • Ensure empty pages aren't generating nulls.
   • The ConditionalSkill should skip null Summary.
6. Tune Indexer parameters & handle failed docs

      "parameters": {
        "maxFailedItems": -1,
        "maxFailedItemsPerBatch": -1
      }
   

   Verify indexer execution history > Docs failed/succeeded via portal.- https://learn.microsoft.com/en-us/azure/search/cognitive-search-concept-troubleshooting and https://github.com/MicrosoftDocs/azure-docs/blob/main/articles/search/cognitive-search-concept-troubleshooting.md
7. Lastly. if incremental enrichment is needed, enable enrichment cache preview and perform on Azure CLI, REST-based validation by using:

      az search skillset show --name mySkillset --service-name ...
      az search indexer run --name myIndexer --service-name ...
   

   Then, check JSON, to ensure mappings follow /document/…/*.

I hope this is helpful! Do not hesitate to let me know if you have any other questions or clarifications.

Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.