h1 (Lean harness)

A simple Lean agent harness that scales proof attempts through execution feedback and self-reflection.

• Developer / Organization: Skander Moalla, Ahmad Rammal, Niket Patel, Remi Munos and Julia Kempe
• Category: Related research systems & hosted services
• Purpose: Study how far a simple Lean agent loop can scale.
• Key features: Efficient Lean interaction, verifier feedback and self-reflection without a prescribed decomposition pipeline; paper link.
• Availability: Official project or research page; access varies
• License: Not verified in the linked source (source (opens in a new tab))
• Website / Source code: https://arxiv.org/abs/2610.35962 (opens in a new tab)
• Version: No public software version found in the linked source
• Version source: Official release or documentation (opens in a new tab)
• Last checked: 2026-11-01