design: add preserved thinking support v3 draft

[jimbrub]

Human Overview

Description

Draft design for keeping Strands context management working on preserved-thinking models (Claude Fable 5.1, Opus 5.5, Sonnet 5.5, Haiku 5.5). These models bind each thinking block to the exact system prompt, tools, and history before it, so any client-side edit (summarization, sliding window, truncation, changing system_prompt) causes a 400, or the thinking blocks are silently dropped when drop_block is on. The default conversation manager's summarizer breaks as a result.

The proposal adds a contextManagedServerSide flag on the model config. When it is set, the summarizer hands reduction to a server-side API (Offload.summarize where available, Anthropic compaction otherwise) instead of rewriting history locally. Part 1 covers the summarizer only; Part 2 (the rest of the ContextManager) and the open questions are still in progress.

Public API Changes

export interface BaseModelConfig {
  contextWindowLimit?: number
  contextManagedServerSide?: boolean
}

Related Issues

Resolves: #4933

Documentation PR

N/A, design doc only.

Type of Change

Documentation update

Testing

Design doc only, no code changes.

• I ran hatch run prepare

Checklist

• I have read the CONTRIBUTING document
• I have reviewed and understand every line of code in this PR, including any generated by AI tools, and I can explain why it works
• My change is focused and reasonably small; I have split unrelated work into separate PRs
• I have added any necessary tests that prove my fix is effective or my feature works
• I have updated the documentation accordingly
• I have added an appropriate example to the documentation to outline the feature, or no new docs are needed
• My changes generate no new warnings
• Any dependent changes have been merged and published

---

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.