2.2.0: prompt-builder settings for prompt mode - #31
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 88c5f3dd29
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 review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if has_target_structure(out, settings): | ||
| return False |
There was a problem hiding this comment.
When generic_chat is selected, a local model can return the original request merely wrapped in the required Markdown headings (for example ## Task/## Context/## Constraints/## Output). This early return treats that as non-weak before the word/line overlap checks run, and the later contract gate only validates section presence, so call_flm emits a near-verbatim echo instead of retrying or falling back.
Useful? React with 👍 / 👎.
2 participants
Summary
prompt:mode while keeping the Claude Code default byte-identicalgeneric_chatMarkdown adapter with target-aware validation, deterministic repair/fallback, and dashboard previewValidation
python -m ruff check scripts tests toolspython -m pytest tests -q(345 passed)--check scripts/ui/web/app.jspython -m py_compile scripts\ffp_prompt_builder.py scripts\ffp_config.py scripts\ffp_llm_client.py scripts\grammar_fix.py scripts\ffp_daemon.py tools\prompt_builder_eval.pypython tools\prompt_builder_eval.py-> claude_code 2/2, generic_chat 10/10qwen3.5:4b-> claude_code 2/2, generic_chat 10/10 (data\benchmarks\prompt_builder_eval_flm_20260709.json, local ignored artifact)