Port upstream #577: stable system-prompt cache boundary

[bruno320x]

Semantic port of spacedriveapp#577 onto the fork's newer segmented prompt engine.

The exact upstream patch only conflicted in the channel template and prompt-engine exports/implementation. Those were reconciled rather than replaced:

• the cache boundary now sits at this fork's explicit instruction/context seam, after stable Task Board/linked-agent guidance and before volatile world-state context;
• Anthropic caches only the stable prefix;
• non-Anthropic providers strip the marker;
• MCP tool names are deterministically sorted;
• upstream Anthropic request tests are retained;
• fork-native tests prove the segmented channel prompt keeps working memory/status on the volatile side and that split/strip helpers are correct.

This preserves the newer prompt block-map architecture while adding the upstream caching invariant.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Advanced

Run ID: 6f9e7c6a-7805-4f49-9e33-843a00af7f82

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
🔒 Security Review	✅ Completed 2026-09-15T18:08:46.273329Z	448d1da	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.