Local chat-API summary provider validation

Context

Following the completion of substrate primitives (#10691) for the Shared Deployment MVP, we need to validate the summarization provider paths for local models. Currently, Gemini 2.5 Flash is the default.

The Problem

The deployment target requires using local chat-API summarization, specifically the Qwen3-8b chat-API path. Without validating this path and documenting the configuration, the MVP is incomplete for operators constrained to local models.

The Architectural Reality

Summarization relies on OpenAiCompatible.generate or similar abstractions. The active provider configuration must be transparent to operators and verifiable at runtime via the healthcheck endpoint.

The Fix

Empirically validate the Qwen3-8b chat-API path for summarization. Document the operator configuration path for swapping out the default Gemini provider for Qwen3-8b. Surface the active summarization provider in the server healthcheck.

Contract Ledger Matrix

Acceptance Criteria

• Validate Qwen3-8b chat-API path for session summarization.
• Document operator config path in SharedDeployment.md.
• Surface active summarization provider in Memory Core healthcheck.
• Ensure summary generation maintains expected structural integrity.

Out of Scope

• Complex prompt engineering or fine-tuning for Qwen3-8b (assume standard prompt works or requires minimal tuning).
• End-to-end performance benchmarking.

Avoided Traps

• Rejected: Hardcoding provider logic. Must rely on configuration-driven provider selection to ensure the substrate remains model-agnostic.

Related

• Parent Epic: #10721

Origin Session ID: 79042442-bebc-431d-8968-8a2e7d7a1151 Retrieval Hint: query_raw_memories(query="Local chat-API summary provider validation Qwen3-8b Gemini")