Skip to content

Persist AutoResearch proof decomposition progress - #209

Merged
FluffyAIcode merged 1 commit into
mainfrom
AgentMemory/autoresearch-durable-decomposition-0720
Jul 20, 2026
Merged

Persist AutoResearch proof decomposition progress#209
FluffyAIcode merged 1 commit into
mainfrom
AgentMemory/autoresearch-durable-decomposition-0720

Conversation

@FluffyAIcode

Copy link
Copy Markdown
Owner

Summary

  • generate machine-readable AutoResearch verdicts in host code instead of relying on model-formatted tail output
  • parse real Markdown Critic verdicts and persist deduplicated missing lemmas as parent-linked leaf obligations
  • preserve completed GAN checkpoints, ledgers, transcripts, and failure rows when candidate evaluation fails or reverts
  • force each strategy and GAN turn to target exactly one unresolved leaf

Test plan

  • python3 -m pytest -q tests/inference_engine/bridge/test_agent_gan_repl.py tests/inference_engine/bench/test_autoresearch_supervisor.py
  • python3 -m py_compile scripts/agent_gan_repl.py autoresearch/prefill/supervisor.py
  • verified Markdown **Status:** and **Missing lemma:** parsing, child deduplication, leaf selection, and host verdict parsing

Made with Cursor

Generate verdicts deterministically from Critic results, turn missing lemmas into deduplicated child obligations, and preserve every completed GAN run even when candidate evaluation fails.

Co-authored-by: Cursor <cursoragent@cursor.com>
@cursor

cursor Bot commented Jul 20, 2026

Copy link
Copy Markdown

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.

@FluffyAIcode
FluffyAIcode merged commit cec1b48 into main Jul 20, 2026
7 checks passed
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