fix(agent): continue after Chinese local-file ack before tool calls

[LeonSGP43]

What does this PR do?

Fixes premature turn completion when the model emits a Chinese "starting now" acknowledgement for a tool-required local file-writing task, but has not called any tool yet.
The agent now treats this as an intermediate acknowledgement and auto-continues so the next call can emit actual tool calls.

Related Issue

Fixes #26892

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)
• ✨ New feature (non-breaking change that adds functionality)
• 🔒 Security fix
• 📝 Documentation update
• ✅ Tests (adding or improving test coverage)
• ♻️ Refactor (no behavior change)
• 🎯 New skill (bundled or hub)

Changes Made

• Expanded AIAgent._looks_like_codex_intermediate_ack(...) in run_agent.py:
  • Added Chinese future-action acknowledgement detection.
  • Added persistent local-file intent markers (保存到本地, 写入文件, save to file, etc.).
  • Added write/save action markers to classify tool-required follow-up turns.
• Added regression test test_run_conversation_codex_continues_after_chinese_ack_for_local_file_write in tests/run_agent/test_run_agent_codex_responses.py.

How to Test

1. uv run --extra dev python -m pytest -q tests/run_agent/test_run_agent_codex_responses.py -k chinese_ack_for_local_file_write
2. uv run --extra dev python -m pytest -q tests/run_agent/test_run_agent_codex_responses.py -k continues_after_ack
3. python3 -m py_compile run_agent.py tests/run_agent/test_run_agent_codex_responses.py

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform: Ubuntu 22.04 (reported issue platform) and local macOS runner

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — or N/A
• I've updated cli-config.yaml.example if I added/changed config keys — or N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — or N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide — or N/A
• I've updated tool descriptions/schemas if I changed tool behavior — or N/A

For New Skills

N/A

Screenshots / Logs

• Targeted regression checks:
  • 1 passed for -k chinese_ack_for_local_file_write
  • 2 passed for -k continues_after_ack

[teknium1]

Thanks for the focused regression case. The underlying CJK failure is still present on current main: agent/agent_runtime_helpers.py:2690-2694 accepts only English future-ack phrases, and agent/conversation_loop.py:5103-5135 relies on that detector before issuing the continuation.

Problems

• The patch targets an implementation that moved. run_agent.py:1554-1565 is now a forwarder; the active detector is agent/agent_runtime_helpers.py:2661-2759 and the loop is agent/conversation_loop.py:5103-5135 (extracted in 053025238).
• The positive-only regression does not protect against broad CJK markers such as 开始 and 写 continuing a conversational or final response. Current detector guardrails are documented in agent/agent_runtime_helpers.py:2670-2679.

Suggested changes

• Salvage the CJK logic into agent/agent_runtime_helpers.py and add positive/negative helper tests in tests/agent/test_intent_ack_continuation.py, preserving the current workspace and config-mode contracts.
• The timeline-linked open CJK PRs (fix(agent): add CJK future-ack and action markers to intent-ack continuation detector #55670, fix(agent): add CJK future-ack patterns for intent-ack continuation #55681, fix(agent): continue on CJK intent acknowledgements #55758) cover the same detector; consolidate the chosen behavior there rather than applying the obsolete run_agent.py diff.

Automated hermes-sweeper review.