SEVerA: Verified Synthesis of Self-Evolving Agents CCS: LLM Agents Automated Verification

Debangshu Banerjee, Changming Xu, Gagandeep Singh · 2026

Read SEVerA: Verified Synthesis of Self-Evolving Agents CCS: LLM Agents Automated Verification by Debangshu Banerjee, Changming Xu, Gagandeep Singh (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-25111.txt.

Description

Banerjee and co-authors add proof to self-improving AI agents with SEVerA. Such agents write their own code and tune it per task, but nothing checks their safety. The team frames agent building as learning under hard rules. Each model call gets a formal output contract in first-order logic. A rejection sampler with a checked fallback guards each call. Search, proof, and learning stages then tune the agent without breaking the rules. Across proof, math, and tool tasks, violations stay at zero while scores beat prior systems.

Catalog record

CREATORDebangshu Banerjee, Changming Xu, Gagandeep Singh
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE165841
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.25111

Browse the full library · About the Tower of Babel