{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:FN2I4DW2LJUQ3Q54DDVOOYSHXN","short_pith_number":"pith:FN2I4DW2","schema_version":"1.0","canonical_sha256":"2b748e0eda5a690dc3bc18eae76247bb60563ac2e6c799e68822ce5b53afc23b","source":{"kind":"arxiv","id":"2603.25723","version":2},"attestation_state":"computed","paper":{"title":"Natural-Language Agent Harnesses","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Hai-Tao Zheng, Jingchen Ni, Lexiao Zou, Linyue Pan, Shuo Guo","submitted_at":"2026-03-26T17:58:15Z","abstract_excerpt":"Agent performance is strongly shaped by the surrounding harness: the external execution system around a model that organizes a task run. Yet this logic is usually buried in tightly coupled controller code, which makes harnesses hard to inspect, compare, transfer, and ablate. This paper asks whether the reusable design pattern of an agent harness can be represented as an executable natural-language object. We introduce Natural-Language Agent Harnesses (NLAHs), editable documents that describe run-level harness policy, and Intelligent Harness Runtime (IHR), a shared runtime that interprets these"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2603.25723","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-03-26T17:58:15Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"58d8793125635f8a5c0732598c78344354468b6c38199f9f4893b3bd26ecba89","abstract_canon_sha256":"23d9308d6798c53a544896376038d4d5ff86231ae2552e074633b9aa048ea8db"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:04:30.138023Z","signature_b64":"rTgtkfQGd7ieXO9KxJb6a1QmCKRO5a9MZhE6mNpnUKjlR2lgejUhGMajWx2D9ePruOa6hoisPeyXCd37RvvDDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2b748e0eda5a690dc3bc18eae76247bb60563ac2e6c799e68822ce5b53afc23b","last_reissued_at":"2026-05-20T00:04:30.137087Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:04:30.137087Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Natural-Language Agent Harnesses","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Hai-Tao Zheng, Jingchen Ni, Lexiao Zou, Linyue Pan, Shuo Guo","submitted_at":"2026-03-26T17:58:15Z","abstract_excerpt":"Agent performance is strongly shaped by the surrounding harness: the external execution system around a model that organizes a task run. Yet this logic is usually buried in tightly coupled controller code, which makes harnesses hard to inspect, compare, transfer, and ablate. This paper asks whether the reusable design pattern of an agent harness can be represented as an executable natural-language object. We introduce Natural-Language Agent Harnesses (NLAHs), editable documents that describe run-level harness policy, and Intelligent Harness Runtime (IHR), a shared runtime that interprets these"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.25723","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2603.25723/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2603.25723","created_at":"2026-05-20T00:04:30.137248+00:00"},{"alias_kind":"arxiv_version","alias_value":"2603.25723v2","created_at":"2026-05-20T00:04:30.137248+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.25723","created_at":"2026-05-20T00:04:30.137248+00:00"},{"alias_kind":"pith_short_12","alias_value":"FN2I4DW2LJUQ","created_at":"2026-05-20T00:04:30.137248+00:00"},{"alias_kind":"pith_short_16","alias_value":"FN2I4DW2LJUQ3Q54","created_at":"2026-05-20T00:04:30.137248+00:00"},{"alias_kind":"pith_short_8","alias_value":"FN2I4DW2","created_at":"2026-05-20T00:04:30.137248+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":14,"internal_anchor_count":14,"sample":[{"citing_arxiv_id":"2605.21516","citing_title":"Harnesses for Inference-Time Alignment over Execution Trajectories","ref_index":52,"is_internal_anchor":true},{"citing_arxiv_id":"2605.18747","citing_title":"Code as Agent Harness","ref_index":176,"is_internal_anchor":true},{"citing_arxiv_id":"2605.18476","citing_title":"AI4BayesCode: From Natural Language Descriptions to Validated Modular Stateful Bayesian Samplers","ref_index":73,"is_internal_anchor":true},{"citing_arxiv_id":"2605.12087","citing_title":"Intermediate Artifacts as First-Class Citizens: A Data Model for Durable Intermediate Artifacts in Agentic Systems","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2605.12239","citing_title":"Harness Engineering as Categorical Architecture","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11665","citing_title":"Nautilus: From One Prompt to Plug-and-Play Robot Learning","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10870","citing_title":"Remember the Decision, Not the Description: A Rate-Distortion Framework for Agent Memory","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"2605.08741","citing_title":"Training with Harnesses: On-Policy Harness Self-Distillation for Complex Reasoning","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10365","citing_title":"Agent-ValueBench: A Comprehensive Benchmark for Evaluating Agent Values","ref_index":43,"is_internal_anchor":true},{"citing_arxiv_id":"2605.09186","citing_title":"Agentic MIP Research: Accelerated Constraint Handler Generation","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2605.06365","citing_title":"From Agent Loops to Deterministic Graphs: Execution Lineage for Reproducible AI-Native Work","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2604.22937","citing_title":"AutoPyVerifier: Learning Compact Executable Verifiers for Large Language Model Outputs","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2604.22748","citing_title":"Agentic World Modeling: Foundations, Capabilities, Laws, and Beyond","ref_index":288,"is_internal_anchor":true},{"citing_arxiv_id":"2604.18292","citing_title":"Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence","ref_index":71,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FN2I4DW2LJUQ3Q54DDVOOYSHXN","json":"https://pith.science/pith/FN2I4DW2LJUQ3Q54DDVOOYSHXN.json","graph_json":"https://pith.science/api/pith-number/FN2I4DW2LJUQ3Q54DDVOOYSHXN/graph.json","events_json":"https://pith.science/api/pith-number/FN2I4DW2LJUQ3Q54DDVOOYSHXN/events.json","paper":"https://pith.science/paper/FN2I4DW2"},"agent_actions":{"view_html":"https://pith.science/pith/FN2I4DW2LJUQ3Q54DDVOOYSHXN","download_json":"https://pith.science/pith/FN2I4DW2LJUQ3Q54DDVOOYSHXN.json","view_paper":"https://pith.science/paper/FN2I4DW2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2603.25723&json=true","fetch_graph":"https://pith.science/api/pith-number/FN2I4DW2LJUQ3Q54DDVOOYSHXN/graph.json","fetch_events":"https://pith.science/api/pith-number/FN2I4DW2LJUQ3Q54DDVOOYSHXN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FN2I4DW2LJUQ3Q54DDVOOYSHXN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FN2I4DW2LJUQ3Q54DDVOOYSHXN/action/storage_attestation","attest_author":"https://pith.science/pith/FN2I4DW2LJUQ3Q54DDVOOYSHXN/action/author_attestation","sign_citation":"https://pith.science/pith/FN2I4DW2LJUQ3Q54DDVOOYSHXN/action/citation_signature","submit_replication":"https://pith.science/pith/FN2I4DW2LJUQ3Q54DDVOOYSHXN/action/replication_record"}},"created_at":"2026-05-20T00:04:30.137248+00:00","updated_at":"2026-05-20T00:04:30.137248+00:00"}