chore: force-sync upstream (main)

[ashsolei]

Force-merge -X theirs + iAiFy overlay (wave4-5 runbook). Pre-sync=4db2d115e8ef02f6a6c81aac3f13b1dbae3b79c8. Fleet sync 2026-07-23.

Summary by CodeRabbit

• Bug Fixes

  • Async callable instructions, handoffs, and tool-use behaviors are now handled reliably.
  • Improved parsing of GNU- and BSD-style device entries.
  • Unsupported audio buffer formats now produce a clear error.
  • Realtime event errors provide safer, more consistent diagnostics.

• Security & Privacy

  • Diagnostic logging now respects data-redaction settings and avoids exposing sensitive details.

• Tests

  • Expanded coverage for async callbacks, redacted logging, device parsing, audio validation, and proxy-isolated test environments.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: d6e0a35a-a417-442e-8f9e-ba8a5c2940a8

📥 Commits

Reviewing files that changed from the base of the PR and between 4db2d11 and 7a466f9.

📒 Files selected for processing (20)

• src/agents/agent.py
• src/agents/logger.py
• src/agents/realtime/agent.py
• src/agents/realtime/handoffs.py
• src/agents/realtime/openai_realtime.py
• src/agents/run_internal/turn_resolution.py
• src/agents/run_state.py
• src/agents/sandbox/util/parse_utils.py
• src/agents/voice/models/openai_stt.py
• tests/conftest.py
• tests/realtime/test_agent.py
• tests/realtime/test_openai_realtime.py
• tests/realtime/test_realtime_handoffs.py
• tests/sandbox/test_parse_utils.py
• tests/test_agent_config.py
• tests/test_agent_instructions_signature.py
• tests/test_error_logging_redaction.py
• tests/test_test_environment.py
• tests/test_tool_use_behavior.py
• tests/voice/test_openai_stt.py

---

📝 Walkthrough

Walkthrough

Callable execution now awaits returned awaitables across agent, realtime handoff, and tool-use paths. Logging gains policy-gated diagnostics, sandbox parsing handles device nodes, audio validation rejects unsupported dtypes, and tests remove ambient proxy settings.

Changes

Async callable execution

Layer / File(s)	Summary
Awaitable-result execution src/agents/agent.py, src/agents/realtime/agent.py, src/agents/realtime/handoffs.py, src/agents/run_internal/turn_resolution.py	Callable results are awaited based on inspect.isawaitable, including instruction, handoff, and tool-use callbacks.
Async callable coverage tests/realtime/*, tests/test_agent_config.py, tests/test_agent_instructions_signature.py, tests/test_tool_use_behavior.py	Tests cover async and synchronous callable objects across instruction, handoff, cloning, and tool-use flows.

Diagnostic logging and redaction

Layer / File(s)	Summary
Policy-gated diagnostic warnings src/agents/logger.py, src/agents/run_state.py	Diagnostic arguments are evaluated lazily and omitted when model or tool data logging is disabled or evaluation fails.
Realtime validation logging src/agents/realtime/openai_realtime.py	Server-event and raw-message failure logs use simplified event extraction and policy-controlled diagnostics.
Redaction and caller-context tests tests/realtime/test_openai_realtime.py, tests/test_error_logging_redaction.py	Tests verify redacted records, preserved diagnostics, fallback behavior, caller locations, and emitted error events.

Sandbox device parsing

Layer / File(s)	Summary
Device-node parsing and tests src/agents/sandbox/util/parse_utils.py, tests/sandbox/test_parse_utils.py	GNU and BSD character/block device layouts are parsed with zero sizes and correct names and attributes.

Audio dtype validation

Layer / File(s)	Summary
Unsupported audio dtype handling src/agents/voice/models/openai_stt.py, tests/voice/test_openai_stt.py	Audio conversion raises UserError for dtypes other than int16 and float32, with tests for int32 and float64.

Test environment isolation

Layer / File(s)	Summary
Ambient proxy cleanup and tests tests/conftest.py, tests/test_test_environment.py	Proxy variables are removed during test setup unless the opt-in variable is enabled, with tests for both paths.

Estimated code review effort: 4 (Complex) | ~45 minutes

Suggested reviewers

Suggested reviewers: seratch

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch chore/force-sync-upstream

---

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.}