perf(claude-agent-sdk): reduce tracing payload copies

[Abhijeet Prasad (AbhiPrasad)]

Summary

• Reuse each AssistantMessage serialization for the new LLM span output and conversation history.
• Defer merged LLM output logging until span finalization to avoid repeated deep copies.
• Store tool dispatch keys at span start and reuse them when closing tool spans.

Validation

• Commit hooks passed, including Ruff format/check and codespell.
• Tests not run.
• No cassette files changed or re-recorded.

[chatgpt-codex-connector]

Codex Review Summary

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

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-15T23:52:11.446839Z	6e61170	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 (@codex) review" or "Codex (@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.

[Abhijeet Prasad (AbhiPrasad)]

released with https://github.com/braintrustdata/braintrust-sdk-python/releases/tag/py-sdk-v0.46.1