{"paper":{"title":"Beyond Skill Evolution: Self-Evolving Context Management Policies for Long-Horizon Agent Harnesses","arxiv_id":"2609.34649","primary_cat":"cs.AI","submitted_at":"2026-09-28T08:57:52+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":0.71,"id":"s0","own":false,"role":"background","text":"Harness evolution improves LLM agents by learning from execution trajectories, but existing experience- and skill-based methods are less effective on long-horizon tasks.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.09,"main_claim_p":0.07,"quantitative":false,"quantitative_p":0.46,"role_probabilities":{"gap":0.29,"method":0.0,"result":0.0,"background":0.71,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":1,"p":0.76,"id":"s1","own":false,"role":"background","text":"As interactions grow, useful evidence can be buried by redundant or outdated context, making context management itself a key bottleneck.","hedge":{"p":0.94,"label":"asserted"},"own_p":0.1,"main_claim_p":0.07,"quantitative":false,"quantitative_p":0.05,"role_probabilities":{"gap":0.24,"method":0.0,"result":0.0,"background":0.76,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":2,"p":0.96,"id":"s2","own":true,"role":"contribution","text":"We introduce ContextEvo, a framework that learns a context policy from long-horizon trajectories.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.92,"main_claim_p":0.12,"quantitative":false,"quantitative_p":0.03,"role_probabilities":{"gap":0.0,"method":0.04,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.96,"interpretation":0.0}},{"i":3,"p":0.99,"id":"s3","own":true,"role":"method","text":"ContextEvo reconstructs the model-visible context at key decision points, identifies context-related failures, and applies targeted policy updates.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.92,"main_claim_p":0.1,"quantitative":false,"quantitative_p":0.03,"role_probabilities":{"gap":0.0,"method":0.99,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.01,"interpretation":0.0}},{"i":4,"p":1.0,"id":"s4","own":true,"role":"result","text":"Starting from the open-source Pi-agent harness, ContextEvo improves performance across three long-horizon task benchmarks, achieving results comparable to or better than several prominent agent harnesses, including Codex, OpenCode, and OpenClaw.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.95,"main_claim_p":0.54,"quantitative":true,"quantitative_p":0.97,"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":5,"p":0.64,"id":"s5","own":true,"role":"result","text":"Additional analyses show that fixed or locally evolved context strategies can fall short under long-horizon information pressure, while our methods adapt to the information demands of each environment.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.88,"main_claim_p":0.28,"quantitative":true,"quantitative_p":0.67,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.64,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.02,"interpretation":0.34}}]},"compiled":{"main_claim":{"p":0.77,"text":"Starting from the open-source Pi-agent harness, ContextEvo improves performance across three long-horizon task benchmarks, achieving results comparable to or better than several prominent agent harnesses, including Codex, OpenCode, and OpenClaw.","location":"abstract s4","hedge_strength":{"p":1.0,"label":"asserted"}},"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":4952,"cost_usd_reported":0.0},"compiled_at":"2026-09-29","bank_version":"abstract-p1-20260923","model_version":"classifier-1.13.0","abstract_states":{"claims_first":0.08,"states_limitation":0.21,"human_participants":0.04,"controlled_comparison":0.76,"randomized_assignment":0.02,"states_quantitative_result":0.09,"compares_with_prior_results":0.97,"mentions_code_or_data_availability":0.1},"evidential_basis":{"p":0.97,"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":0.97,"simulation_or_computation":0.03}},"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":0.97,"text":"We introduce ContextEvo, a framework that learns a context policy from long-horizon trajectories.","location":"abstract s2"}},"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":6,"sentences_in_abstract":6},"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"}}}