Skip to content

feat(planner): adopt native Pydantic AI runtime - #196

Open
wilburx813 wants to merge 11 commits into
RLinf:mainfrom
wilburx813:feat/api-harness
Open

wilburx813 wants to merge 11 commits into
RLinf:mainfrom
wilburx813:feat/api-harness

Conversation

@wilburx813

Copy link
Copy Markdown
Collaborator

Description

  • Replace the custom API planner loop with Pydantic AI's native runtime.
  • Integrate the native CLI for terminal interaction and adapt Dashboard
    steering and cancellation.
  • Use Harness sliding-window trimming to manage conversation history.
  • Preserve tool history after failures, support finish retries, and
    retain Anthropic prompt caching.
  • Update dependencies, regression tests, and related documentation.

Testing

Local checks with Python 3.12:

  • pre-commit run --all-files: passed.
  • python -m pytest tests/unit_tests -q -ra: 598 passed, 3 skipped
    due to missing optional RLinf and LeRobot dependencies.
  • sphinx-build -W --keep-going docs/source-en docs/build/html-en: passed.
  • sphinx-build -W --keep-going docs/source-zh docs/build/html-zh: passed.
  • python -m rpent.cli.main --help, API planner import, and pip check: passed.

Manual verification

No live model API, simulator, or real-robot verification performed.

Checklist

  • My code follows the code style of this project.
  • I have updated related documentation when needed.
  • I have added tests to cover my changes when needed.
  • All new and existing tests passed.

Comment thread rpent/planner/api_loop.py Outdated
Comment thread rpent/planner/api_loop.py
@qurakchin
qurakchin self-requested a review September 17, 2026 07:49
Comment thread docs/source-zh/rst_source/usage/configure_planner.rst Outdated
Comment thread docs/source-zh/rst_source/usage/configure_planner.rst Outdated
Comment thread rpent/cli/main.py
@qurakchin
qurakchin self-requested a review September 22, 2026 03:20
Comment thread rpent/planner/api_loop.py
@wilburx813

wilburx813 commented Sep 24, 2026 •

Copy link
Copy Markdown
Collaborator Author

The API planner achieved 43.3% success on LIBERO-PRO Long Swap (13/30 cells) using GPT-5.6 Sol with no reasoning. exactly same with using codex planner with gpt5.6-sol no reasoning.

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.

2 participants