{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:MZPCQMLFE2S6KICUVDQ4JCEFLC","short_pith_number":"pith:MZPCQMLF","schema_version":"1.0","canonical_sha256":"665e28316526a5e52054a8e1c4888558aa0070360123dfcd1f17088301f4376d","source":{"kind":"arxiv","id":"2607.05458","version":1},"attestation_state":"computed","paper":{"title":"Learning to Control LLM Agent Harnesses with Offline Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Haiwen Yi, Xinyuan Song","submitted_at":"2026-07-05T22:11:18Z","abstract_excerpt":"Large language model (LLM) agents are usually improved by changing prompts, models, or hand-written workflows, while the execution harness around the model is treated as fixed infrastructure. We argue that this harness is itself a learnable control layer. We formalize harness operation as a finite-horizon Harness MDP, where a lightweight controller selects structural execution actions while the LLM executor remains frozen. The controller is trained from offline rollouts using advantage-weighted regression with only terminal task-rubric rewards. We also separate final task quality from a post-h"},"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":"2607.05458","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-07-05T22:11:18Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"54321d93674921a54662265b1e6818472702832b8f7013dfb4ece3d141221099","abstract_canon_sha256":"82aab22d72e9dd5df73e46a78be8168c35a1af08450b1bc7682edf5bdabf97a7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-08T01:18:10.545606Z","signature_b64":"/NSP7n3r23hqQKGptAQnNO2w4udC5J9UVOK/WO404GOB4D493DSCUkUiLI1+1X7wmMp+7XyY63TrSyh70ED6DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"665e28316526a5e52054a8e1c4888558aa0070360123dfcd1f17088301f4376d","last_reissued_at":"2026-07-08T01:18:10.545174Z","signature_status":"signed_v1","first_computed_at":"2026-07-08T01:18:10.545174Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning to Control LLM Agent Harnesses with Offline Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Haiwen Yi, Xinyuan Song","submitted_at":"2026-07-05T22:11:18Z","abstract_excerpt":"Large language model (LLM) agents are usually improved by changing prompts, models, or hand-written workflows, while the execution harness around the model is treated as fixed infrastructure. We argue that this harness is itself a learnable control layer. We formalize harness operation as a finite-horizon Harness MDP, where a lightweight controller selects structural execution actions while the LLM executor remains frozen. The controller is trained from offline rollouts using advantage-weighted regression with only terminal task-rubric rewards. We also separate final task quality from a post-h"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.05458","kind":"arxiv","version":1},"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/2607.05458/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":"2607.05458","created_at":"2026-07-08T01:18:10.545224+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.05458v1","created_at":"2026-07-08T01:18:10.545224+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.05458","created_at":"2026-07-08T01:18:10.545224+00:00"},{"alias_kind":"pith_short_12","alias_value":"MZPCQMLFE2S6","created_at":"2026-07-08T01:18:10.545224+00:00"},{"alias_kind":"pith_short_16","alias_value":"MZPCQMLFE2S6KICU","created_at":"2026-07-08T01:18:10.545224+00:00"},{"alias_kind":"pith_short_8","alias_value":"MZPCQMLF","created_at":"2026-07-08T01:18:10.545224+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MZPCQMLFE2S6KICUVDQ4JCEFLC","json":"https://pith.science/pith/MZPCQMLFE2S6KICUVDQ4JCEFLC.json","graph_json":"https://pith.science/api/pith-number/MZPCQMLFE2S6KICUVDQ4JCEFLC/graph.json","events_json":"https://pith.science/api/pith-number/MZPCQMLFE2S6KICUVDQ4JCEFLC/events.json","paper":"https://pith.science/paper/MZPCQMLF"},"agent_actions":{"view_html":"https://pith.science/pith/MZPCQMLFE2S6KICUVDQ4JCEFLC","download_json":"https://pith.science/pith/MZPCQMLFE2S6KICUVDQ4JCEFLC.json","view_paper":"https://pith.science/paper/MZPCQMLF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.05458&json=true","fetch_graph":"https://pith.science/api/pith-number/MZPCQMLFE2S6KICUVDQ4JCEFLC/graph.json","fetch_events":"https://pith.science/api/pith-number/MZPCQMLFE2S6KICUVDQ4JCEFLC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MZPCQMLFE2S6KICUVDQ4JCEFLC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MZPCQMLFE2S6KICUVDQ4JCEFLC/action/storage_attestation","attest_author":"https://pith.science/pith/MZPCQMLFE2S6KICUVDQ4JCEFLC/action/author_attestation","sign_citation":"https://pith.science/pith/MZPCQMLFE2S6KICUVDQ4JCEFLC/action/citation_signature","submit_replication":"https://pith.science/pith/MZPCQMLFE2S6KICUVDQ4JCEFLC/action/replication_record"}},"created_at":"2026-07-08T01:18:10.545224+00:00","updated_at":"2026-07-08T01:18:10.545224+00:00"}