What problem does it solve? Building HTTP-facing AI agent endpoints requires coordinating agent runners, session memory, dependency injection, and error handling, and this Skill provides the canonical patterns for wiring Google ADK agents into Litestar applications without exposing agent internals as the API contract. ## Core Features & Use Cases - ADK Provider Wiring: Configure LlmAgent, Runner, and SQLSpecSessionService through Dishka providers with correct APP and REQUEST scopes. - Session-Backed Chat Handlers: Implement request/response chat endpoints that persist multi-turn conversation state across HTTP requests via session IDs. - Credential and Failure Guards: Convert missing Vertex AI credentials into HTTP 503 responses and enforce tool-use-first prompt patterns to prevent hallucinated answers. - Use Case: You are building a support assistant API where users send chat messages, the agent classifies intent, calls search tools, and remembers prior turns; this Skill gives you the controller, runner, and DI structure to ship it. ## Quick Start Use the litestar-ai-serving skill to create a Litestar chat endpoint backed by a Google ADK agent with persistent session memory.
