Known Issues and Bugs

This section describes open bugs and known issues for Agent Factory release 25.3.0.0.8.

Known Issues Affecting Knowledge Agent

Bug 39059407

In Knowledge Agent, if a generative model configuration using a model from Ollama, vLLM, or Gemini is set as default, and then a different generative model configuration (also from Ollama, vLLM, or Gemini) is subsequently set as default, only the first generative model configuration set in this sequence will continue to be used by the Knowledge Agent.

A similar issue may occur for embedding models from the Ollama or vLLM providers.

Workaround:

Restart the knowledge agent using the following steps:

1. Enter the podman container.

2. Run the following commands:

   cd /home/aaiuser/install/agent_factory_components/knowledge_agent/
   ./manage.sh restart
   

Bug 39077443

If an embedding model configuration from Ollama, vLLM, OpenAI, or Gemini encounters an error while embedding a data source, the embedding process will enter an infinite loop.

Workaround:

Ensure that the default embedding model configuration is functional. Then, restart the Knowledge Agent using the following steps:

1. Enter the podman container.

2. Run the following commands:

   cd /home/aaiuser/install/agent_factory_components/knowledge_agent/
   ./manage.sh restart
   

Bug 38933173

In the Knowledge Agent, ingestion may fail with the error “No extractable text content found in data source (‘ZERO PARSED UNITS’)”. This occurs when the files in the specified data source cannot be processed because none of them match a supported modality or format for content extraction. This situation can arise if the files are empty, contain only non-extractable content (e.g., scanned images without OCR), or use an unsupported encoding or format.

Note: Specific files within a data source may fail ingestion even if others succeed. If expected content is missing, check the alert.log to verify whether those files were skipped or failed during processing.

Workaround:

• For File System data sources:  - Verify that the data source contains supported documents with extractable text content.
  • Ensure files are not empty and are in a supported format.
• For Web data sources:  No direct workaround is available. Use an alternative web source configuration that provides extractable content.
• SharePoint / Google Drive data sources: If you have access to the underlying workspace, verify that the files are in supported formats, contain extractable text (not just scanned images without OCR), and are not empty or corrupted.