--- format: "aidr-story-markdown/v1" id: "4c6e95532e0e57579af3457c9f839143118811919a86e1633a6722b6a861705c" canonical_url: "https://aidr.today/4c6e9553?lang=en" title: "Harness^2: Recursive Agent Harnessing for an Open World" lang: "en" requested_lang: "en" available_langs: ["en","vi"] translation_fallback: null fallback_fields: [] published_at: "2026-10-14T00:00:00.000Z" category: "Frameworks" topics: ["agent","framework"] source_urls: ["https://huggingface.co/papers/2610.14224","https://huggingface.co/papers/2610.18356"] summary: "Researchers have introduced Harness^2, a plug-and-play recursive agent harnessing framework designed to address challenges in open-world settings. The framework uses task-level recursion with a contrastive Propose-Probe-Compose step to refine harnesses without ground-truth verifiers, and harnessing-level recursion to extract experience for continual improvement across future tasks." --- # Harness^2: Recursive Agent Harnessing for an Open World > [Open the canonical story]() **Published:** 2026-10-14T00:00:00.000Z **Category:** Frameworks **Topics:** agent, framework ## Summary Researchers have introduced Harness^2, a plug\-and\-play recursive agent harnessing framework designed to address challenges in open\-world settings\. The framework uses task\-level recursion with a contrastive Propose\-Probe\-Compose step to refine harnesses without ground\-truth verifiers, and harnessing\-level recursion to extract experience for continual improvement across future tasks\. ## Sources - [Story source]() - [Story source]()