Understand this story
On November 16, 2026, Pydantic AI Harness launched the AzureDurableTaskDurability capability, executing model requests, tool calls, and MCP calls all as Azure Durable Task activities. According to the official Pydantic AI article, after a worker crashes, a new worker replays historical results and continues from the interrupted step. This capability is based on Microsoft's Durable Task engine, can be debugged locally using a Docker emulator, and can also be deployed to Azure Durable Task Scheduler. The official article also states that activities execute at least once, tools need to use ctx.run_id and ctx.tool_call_id for idempotency, and orchestrator-side code reruns on every replay. There are currently no subsequent development facts.
AI-generated from public reports
Latest development · Report
Pydantic AI Harness introduces AzureDurableTaskDurability, enabling agents to…Developments
1 confirmed developments
- Source:English text is being prepared.
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 16
Official Pydantic AI articleFirst-partySelected
Pydantic AI Harness introduces AzureDurableTaskDurability, enabling agents to…
Pydantic AI Harness adds the AzureDurableTaskDurability capability, executing model requests, tool calls, and MCP calls as Azure Durable Task activities; after a worker crashes, a new worker replays historical results and continues from the interrupted step. This capability is based on Microsoft's Durable Task engine, can be debugged locally with a Docker emulator, and can also be deployed to Azure Durable Task Scheduler. The original article also explains that activities execute at least once, tools must use ctx.run_id and ctx.tool_call_id for idempotency, and orchestrator-side code reruns on every replay.