Agent Framework Foundry Agent error in deploying to Azure durable function - agent not found

Question

Agent Framework Foundry Agent error in deploying to Azure durable function - agent not found

Jaydev 16

As part of MS Agent Framework, on registering a FoundryAgent with a name and version number in a local Python durable function app in VS Code using Azure Function Tools, I get the error on invoking the orchestration trigger with get_agent passing the registered agent name that "Could not find agent with name and version number". Firstly is FoundryAgent supported in Azure Durable function app and if so, why am i getting this error. Thanks

Pravallika KV 13,545 Reputation points Microsoft External Staff Moderator

2026-04-09T06:53:49.27+00:00
Hi @Jaydev,

Thanks for reaching out to Microsoft Q&A.

FoundryAgent is supported in Azure Functions including Durable Functions, but there are a few moving parts you need to line up before get_agent(name, version) will work:

Pre-create or register your agent in your Azure AI Foundry project:

The Durable Function orchestration runs in its own context and won’t “see” an in-memory agent you registered at startup unless that agent is actually deployed in your AI project. Use the Python client to call client.create_agent(...) (or the REST API) and wait until provisioningState == 'Succeeded' before you invoke get_agent.

Double-check your local.settings.json & environment variables In your Function App folder you should have something like:
{ "IsEncrypted": false, "Values": { "AzureWebJobsStorage": "UseDevelopmentStorage=true", "FUNCTIONS_WORKER_RUNTIME": "python", "STORAGE_CONNECTION__queueServiceUri": "https://<yourStorage>.queue.core.windows.net", "PROJECT_CONNECTION_STRING": "<your AI Project connection string>" } }

PROJECT_CONNECTION_STRING must point to the same project where your agent was created.

Durable Functions will spin up separate workers, so each needs those env vars to locate the AI Project and agent registry.

Match agent name & version exactly

get_agent("MyAgent", "1.0.0") will only find an agent you created as name="MyAgent" with version="1.0.0".

Typos or mismatched casing will lead to “Could not find agent” errors.

Use the supported client versions & tooling
azure-ai-projects Python library >= 2.0.0 Azure Functions Core Tools v4+

If after those checks you still see the error, it usually means the agent wasn’t fully provisioned or the function host can’t reach your AI project.

References:

Use Azure Functions with Foundry Agent Service
Anshika Varshney 9,740 Reputation points Microsoft External Staff Moderator

2026-04-13T08:43:26.8133333+00:00

Hi Jaydev,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thankyou!

1 answer

Your answer

Pravallika KV 13,545 Reputation points Microsoft External Staff Moderator

2026-04-09T06:53:49.27+00:00

Hi @Jaydev,

Thanks for reaching out to Microsoft Q&A.

FoundryAgent is supported in Azure Functions including Durable Functions, but there are a few moving parts you need to line up before get_agent(name, version) will work:

Pre-create or register your agent in your Azure AI Foundry project:

The Durable Function orchestration runs in its own context and won’t “see” an in-memory agent you registered at startup unless that agent is actually deployed in your AI project. Use the Python client to call client.create_agent(...) (or the REST API) and wait until provisioningState == 'Succeeded' before you invoke get_agent.

Double-check your local.settings.json & environment variables In your Function App folder you should have something like:
{ "IsEncrypted": false, "Values": { "AzureWebJobsStorage": "UseDevelopmentStorage=true", "FUNCTIONS_WORKER_RUNTIME": "python", "STORAGE_CONNECTION__queueServiceUri": "https://<yourStorage>.queue.core.windows.net", "PROJECT_CONNECTION_STRING": "<your AI Project connection string>" } }

PROJECT_CONNECTION_STRING must point to the same project where your agent was created.

Durable Functions will spin up separate workers, so each needs those env vars to locate the AI Project and agent registry.

Match agent name & version exactly

get_agent("MyAgent", "1.0.0") will only find an agent you created as name="MyAgent" with version="1.0.0".

Typos or mismatched casing will lead to “Could not find agent” errors.

Use the supported client versions & tooling
azure-ai-projects Python library >= 2.0.0 Azure Functions Core Tools v4+

If after those checks you still see the error, it usually means the agent wasn’t fully provisioned or the function host can’t reach your AI project.

References:

Use Azure Functions with Foundry Agent Service
Anshika Varshney 9,740 Reputation points Microsoft External Staff Moderator

2026-04-13T08:43:26.8133333+00:00

Hi Jaydev,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thankyou!

Answer 1

Hi Jaydev,

If you are seeing an error while deploying an agent in Azure AI Foundry using the Agent Framework, this usually happens due to configuration or resource setup issues rather than a platform problem.

Here are a few common things you can check.

First, confirm that your Foundry project and agent service are in a supported region. Agent deployment depends on both the project region and the model region. If the selected model or agent capability is not supported in that region, the deployment may fail.

Second, check the model deployment used by the agent. Agents need an active and healthy model deployment to function. If the selected model is not deployed, failed earlier, or still provisioning, the agent deployment will not complete successfully.

Third, verify identity and permissions. Agent Framework uses managed identity to access resources like model deployments or tools. If the required permissions are not assigned to the project identity, the agent may fail during deployment.

Fourth, confirm tool or resource configuration. If your agent uses tools such as Azure AI Search or storage connections, make sure those resources are properly linked and accessible from the Foundry project. Incorrect or missing connections can block deployment.

Fifth, check quota and deployment limits. Some agent related models or capabilities require available quota. If quota is zero or already fully used, the agent deployment may fail even if access is approved.

To narrow down the issue, you can also try deploying a basic agent without tools first. If that works, then the issue may be related to tool configuration rather than the agent setup itself.

Official references that may help.

Model deployment in Azure AI Foundry https://learn.microsoft.com/azure/ai-studio/how-to/deploy-models

Authentication for Azure AI services https://learn.microsoft.com/azure/ai-services/authentication

In most cases, once the project region, model deployment, identity permissions, and tool connections are aligned, the agent deployment works as expected.

Hope this helps, and feel free to share what you observe after checking these points.

Thankyou!

Share via

Agent Framework Foundry Agent error in deploying to Azure durable function - agent not found

1 answer

Your answer