v0.13.1

Highlights

A small maintenance release on top of v0.13.0. The headline is the Agent Harness Platform — a new flagship, runnable example with a stage-by-stage docs tour showing platform engineers how to build an internal agent platform on Kitaru + PydanticAI. This release also fixes PydanticAI adapter compatibility with the latest pydantic-ai-slim, and bundles the latest Kitaru UI (kitaru-ui-v0.1.10).

Added

• Agent Harness Platform — a chapter-by-chapter flagship example at examples/end_to_end/agent_harness_platform/ with a dedicated docs tour at kitaru.ai/docs/agent-harness-platform. A platform engineer's starter kit for building internal agent platforms on Kitaru + PydanticAI: six runnable stages progress from a 30-line durable agent to a sandboxed, credential-isolated agent with human-in-the-loop, each stage adding exactly one capability — durable execution → command sandbox → markdown skills → credential proxy → typed service calls → HITL via kitaru.wait(). (#288)

Changed

• The examples documentation is reorganized into three categories — Agent Harness Platform tour, other end-to-end examples, and feature-focused examples — replacing the previous goal-keyed table. The news_scout example stays runnable and now lives under "Other end-to-end examples"; the guides section is reserved for Kitaru-feature how-tos. (#288)

Fixed

• Fixed PydanticAI adapter compatibility with pydantic-ai-slim>=1.95, where upstream renamed built-in tools to native tools. The adapter no longer fails at import time on AgentBuiltinTool or crashes by forwarding builtin_tools=None into PydanticAI's deprecation shim. (#370)

Full Changelog: v0.13.0...v0.13.1