{"paper":{"title":"ModHarness-RL: Recasting Harnesses as Learnable and Evolving Programs for Agentic RL","arxiv_id":"2611.18533","primary_cat":"cs.AI","submitted_at":"2026-11-13T18:45:49+00:00","version_compiled":"v1"},"notice":"Every quoted sentence is the authors' text from the arXiv listing abstract. Labels and probabilities are the outputs of classifier-1.13.0 on the question bank abstract-p1-20260923, exactly as returned. No sentence in this record was written by a language model.","schema":"pith.compiled-record/v0","status":"compiled","abstract":{"source":"arXiv listing abstract v1","sentences":[{"i":0,"p":1.0,"id":"s0","own":false,"role":"background","text":"Agentic reinforcement learning (RL) increasingly relies on a harness, the scaffold around the model that decides what it sees of its own interaction and what guidance it receives.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.05,"main_claim_p":0.02,"quantitative":false,"quantitative_p":0.03,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":1.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":1,"p":0.85,"id":"s1","own":false,"role":"background","text":"Most existing approaches train under a hand-written harness that stays fixed throughout.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.05,"main_claim_p":0.02,"quantitative":false,"quantitative_p":0.16,"role_probabilities":{"gap":0.15,"method":0.0,"result":0.0,"background":0.85,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":2,"p":0.62,"id":"s2","own":false,"role":"gap","text":"However, an effective harness depends on the domain, the data and the model, and keeps changing as the model trains.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.08,"main_claim_p":0.06,"quantitative":false,"quantitative_p":0.04,"role_probabilities":{"gap":0.62,"method":0.0,"result":0.0,"background":0.38,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":3,"p":0.99,"id":"s3","own":true,"role":"contribution","text":"We introduce ModHarness-RL, which recasts the harness as programs that are learned from data and evolve as the model trains with RL.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.9,"main_claim_p":0.15,"quantitative":false,"quantitative_p":0.04,"role_probabilities":{"gap":0.0,"method":0.01,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.99,"interpretation":0.0}},{"i":4,"p":0.97,"id":"s4","own":true,"role":"method","text":"ModHarness-RL records the interaction as structured State, a set of typed blocks, and two core components of the harness are programs over these blocks, a Compression program that assembles the accumulated history into the model's context and a Steering program that supplies guidance when behavior deviates.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.82,"main_claim_p":0.04,"quantitative":false,"quantitative_p":0.21,"role_probabilities":{"gap":0.0,"method":0.97,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.03,"interpretation":0.0}},{"i":5,"p":1.0,"id":"s5","own":true,"role":"method","text":"During RL training, candidate programs are proposed from recent trajectories, verified by execution, and replayed over past trajectories, so the harness tracks how the model's behavior shifts.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.89,"main_claim_p":0.07,"quantitative":false,"quantitative_p":0.05,"role_probabilities":{"gap":0.0,"method":1.0,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":6,"p":0.7,"id":"s6","own":true,"role":"method","text":"We study CodeSearch in depth and further analyze ALFWorld and BIRD SQL, with models from 1.5B to 8B.","hedge":{"p":0.9,"label":"asserted"},"own_p":0.92,"main_claim_p":0.02,"quantitative":true,"quantitative_p":0.94,"role_probabilities":{"gap":0.0,"method":0.7,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.3,"interpretation":0.0}},{"i":7,"p":1.0,"id":"s7","own":true,"role":"result","text":"The proposed Compression and Steering mechanisms improve held-out performance over the fixed-harness baseline in most settings.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.95,"main_claim_p":0.23,"quantitative":true,"quantitative_p":0.91,"role_probabilities":{"gap":0.0,"method":0.0,"result":1.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":8,"p":1.0,"id":"s8","own":true,"role":"interpretation","text":"These results suggest that a harness for agentic RL can be learned from the task's data and evolved with the model.","hedge":{"p":1.0,"label":"hedged_possibility"},"own_p":0.77,"main_claim_p":0.9,"quantitative":false,"quantitative_p":0.05,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":1.0}}]},"compiled":{"main_claim":{"p":1.0,"text":"These results suggest that a harness for agentic RL can be learned from the task's data and evolved with the model.","location":"abstract s8","hedge_strength":{"p":1.0,"label":"hedged possibility"}},"provenance":{"route":{"via":"TypeSafe API","model":"classifier-1.13.0","endpoint":"https://api.typesafe.ai/v1/systemone","provider":"TypeSafe"},"packing":"one request per paper","requests":1,"input_tokens":6922,"cost_usd_reported":0.0},"compiled_at":"2026-11-16","bank_version":"abstract-p1-20260923","model_version":"classifier-1.13.0","abstract_states":{"claims_first":0.09,"states_limitation":0.14,"human_participants":0.03,"controlled_comparison":0.96,"randomized_assignment":0.01,"states_quantitative_result":0.19,"compares_with_prior_results":0.96,"mentions_code_or_data_availability":0.04},"evidential_basis":{"p":0.94,"value":"observation or experiment","probabilities":{"argument":0.0,"literature":0.0,"proof_or_derivation":0.0,"analysis_of_existing_data":0.0,"observation_or_experiment":0.94,"simulation_or_computation":0.06}},"contribution_type":{"p":1.0,"value":"method or tool","probabilities":{"review":0.0,"method_or_tool":1.0,"empirical_study":0.0,"simulation_study":0.0,"theoretical_result":0.0,"dataset_or_benchmark":0.0,"position_or_commentary":0.0}},"contribution_statement":{"p":1.0,"text":"We introduce ModHarness-RL, which recasts the harness as programs that are learned from data and evolve as the model trains with RL.","location":"abstract s3"}},"coverage":{"tier":"abstract","levels_run":["P abstract"],"not_examined":["full text","figures and tables","references and citing sentences","relations to other papers"],"sources_read":["arXiv listing abstract v1"],"levels_not_run":["S sections","T body sentences","C citations","R cross-paper relations"],"sentences_examined":9,"sentences_in_abstract":9},"measured_accuracy":{"url":"https://pith.science/accuracy","measured_on":"2026-10-02","method":"two independent model families labelled a fresh random sample blind; the figure is how often the published label matches their consensus","sentence_role_agreement":0.92,"main_claim_agreement":0.908,"typed_citation_agreement":0.933,"typed_citation_agreement_95_interval":[0.917,0.945],"review_comments_accurate":0.758,"review_comments_accurate_95_interval":[0.684,0.82],"review_caught_admitted_error_found_or_related":0.448,"review_comments_judged":{"judged_by":["gpt-5.6-luna","grok-4.7"],"judged_against":"Pith's own text copy of each paper: text extracted from the PDF, the first 90,000 characters","measured_on":"2026-10-04","accurate":113,"decided":149,"sampled_reviews":200,"split_by_lane":false,"by_lane":"https://pith.science/accuracy#by-lane"}}}