[devin-ai-integration]

Summary

Closes a code-scan coverage gap: nothing in tests/ exercised the Anthropic prompt-caching flags, so apply_cache_breakpoints, mark_all_previous_msg_for_caching, and the cache_control injections in ClaudeResponseModel._call_api were entirely untested.

Tests only; no production code changed. Each Claude test patches client.messages.create and asserts on the captured api_params:

• default payload puts no cache_control on tools or message content
• cache_tool_definition=True marks tools[-1] only
• cache_complete_prompt=True marks messages[-1]["content"][-1] only
• use_cache_breakpoints=True marks exactly the builders whose mark_all_previous_msg_for_caching() was called
• pins current behavior for the risky index: a marked message with empty content raises IndexError (documents the crash rather than silently accepting it)
• builder-level: Anthropic builder sets _cache_breakpoint, both OpenAI builders no-op, base MessageBuilder raises NotImplementedError

Devin-Org: engineering

Link to Devin session: https://app.devin.ai/sessions/71d38fe07b4749a09f2de8349f9bcd7c
Open in Devin Desktop: https://app.devin.ai/desktop/session/71d38fe07b4749a09f2de8349f9bcd7c?variant=devin

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring