Skip to content

fix: remove stale workspace-mode prompts from persistent_browser.yaml#46

Open
nuthalapativarun wants to merge 1 commit into
microsoft:mainfrom
nuthalapativarun:fix/persistent-browser-prompt-cleanup
Open

fix: remove stale workspace-mode prompts from persistent_browser.yaml#46
nuthalapativarun wants to merge 1 commit into
microsoft:mainfrom
nuthalapativarun:fix/persistent-browser-prompt-cleanup

Conversation

@nuthalapativarun

Copy link
Copy Markdown

Summary

  • Replaced bare final_script.py reference in Helpful Command Patterns with the correct run-folder path final_runs/run_<id>/final_script.py
  • Replaced hardcoded final_runs/run_003/ paths in image_qa examples with generic run_<id> placeholders
  • Replaced oversimplified done-gate constraint ("final_script.py is the final artifact") with a pointer to the full Completion Gate, including self_reflection and Chromium-release requirements
  • Added summary_user_prompt override referencing judge_config.json instead of the default self_reflect_config.json inherited from base.yaml

Why

PR #16 fixed local_browser.yaml and called out that persistent_browser.yaml had the same copy-pasted workspace-mode prompts that don't accurately reflect the execution model. Specifically:

  • The bare final_script.py inspect pattern implied the script lives at the workspace root, but in this mode it lives under final_runs/run_<id>/
  • The image_qa examples had hardcoded run_003 paths, making them misleading for other run IDs
  • The done-gate global constraint only mentioned final_script.py as the final artifact, omitting the required self_reflection pass and Chromium release steps
  • The inherited default summary_user_prompt references self_reflect_config.json but this mode uses judge_config.json

Changes

  • src/webwright/config/persistent_browser.yaml: prompt cleanup as described above

Testing

pytest tests/ -v
============================= test session starts =============================
platform win32 -- Python 3.14.3, pytest-7.4.4, pluggy-1.6.0
collected 7 items

tests/unit/test_tool_model_routing.py::test_model_claude_sets_top_level_anthropic_model PASSED
tests/unit/test_tool_model_routing.py::test_model_claude_does_not_declare_per_tool_overrides PASSED
tests/unit/test_tool_model_routing.py::test_extract_model_block_reads_top_level_model PASSED
tests/unit/test_tool_model_routing.py::test_extract_model_block_rejects_missing_block PASSED
tests/unit/test_tool_model_routing.py::test_resolve_model_config_path_prefers_explicit_arg PASSED
tests/unit/test_tool_model_routing.py::test_resolve_model_config_path_falls_back_to_workspace_snapshot PASSED
tests/unit/test_tool_model_routing.py::test_resolve_model_config_path_raises_when_nothing_found PASSED

7 passed, 238 warnings in 0.12s

- Replace bare `final_script.py` reference in Helpful Command Patterns
  with the correct run-folder path `final_runs/run_<id>/final_script.py`
- Replace hardcoded `final_runs/run_003/` image_qa examples with generic
  `run_<id>` placeholders
- Replace oversimplified done-gate constraint (\"final_script.py is the
  final artifact\") with a summary that points to the full Completion Gate
- Add summary_user_prompt override to reference judge_config.json instead
  of the default self_reflect_config.json from base.yaml

Follow-up to microsoft#16 which noted the same copy-pasted issues in this file.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant