{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:XLL44TIX7ZIGRVEGOVM3B7QDPF","short_pith_number":"pith:XLL44TIX","schema_version":"1.0","canonical_sha256":"bad7ce4d17fe5068d4867559b0fe037946044f942b257ddd3fa7c94252a00e8f","source":{"kind":"arxiv","id":"2607.20668","version":1},"attestation_state":"computed","paper":{"title":"From Agent Failures to Text Policies: What Works and What Breaks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ankit Goyal, Jaideep Ray","submitted_at":"2026-07-22T19:08:19Z","abstract_excerpt":"TextGrad improves language-model systems by revising text from feedback. Its core thesis is that natural-language feedback can act as a gradient for optimizing text components without changing model weights. Applying it to agents is harder because feedback arrives only after a sequence of actions, making it difficult to identify which decision caused failure. We study this problem by separating the ability to follow a useful policy from the ability to learn that policy from experience. Our main finding is a clear gap between these two abilities. Human-written policies improve two frozen 7B age"},"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":"2607.20668","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-22T19:08:19Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2a3e164db69e8d6bed77422a4e0f05204fa731e28ce06b6cf96781dc9ca8f23e","abstract_canon_sha256":"58a3135f2d782edd0ae9da17b3759fdb839e2bb2031fcbf0a7429b6f4d6f839d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-24T00:23:28.515126Z","signature_b64":"7oi3AtpPlm4n96D0aJDqN7YuP2AQlls/Ba7Dfac5qo5ClORK2LAlbTmu2gM670+XA4svLb4EsJ9AgMq0ElfeDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bad7ce4d17fe5068d4867559b0fe037946044f942b257ddd3fa7c94252a00e8f","last_reissued_at":"2026-07-24T00:23:28.514254Z","signature_status":"signed_v1","first_computed_at":"2026-07-24T00:23:28.514254Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"From Agent Failures to Text Policies: What Works and What Breaks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ankit Goyal, Jaideep Ray","submitted_at":"2026-07-22T19:08:19Z","abstract_excerpt":"TextGrad improves language-model systems by revising text from feedback. Its core thesis is that natural-language feedback can act as a gradient for optimizing text components without changing model weights. Applying it to agents is harder because feedback arrives only after a sequence of actions, making it difficult to identify which decision caused failure. We study this problem by separating the ability to follow a useful policy from the ability to learn that policy from experience. Our main finding is a clear gap between these two abilities. Human-written policies improve two frozen 7B age"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.20668","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/2607.20668/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":"2607.20668","created_at":"2026-07-24T00:23:28.514706+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.20668v1","created_at":"2026-07-24T00:23:28.514706+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.20668","created_at":"2026-07-24T00:23:28.514706+00:00"},{"alias_kind":"pith_short_12","alias_value":"XLL44TIX7ZIG","created_at":"2026-07-24T00:23:28.514706+00:00"},{"alias_kind":"pith_short_16","alias_value":"XLL44TIX7ZIGRVEG","created_at":"2026-07-24T00:23:28.514706+00:00"},{"alias_kind":"pith_short_8","alias_value":"XLL44TIX","created_at":"2026-07-24T00:23:28.514706+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/XLL44TIX7ZIGRVEGOVM3B7QDPF","json":"https://pith.science/pith/XLL44TIX7ZIGRVEGOVM3B7QDPF.json","graph_json":"https://pith.science/api/pith-number/XLL44TIX7ZIGRVEGOVM3B7QDPF/graph.json","events_json":"https://pith.science/api/pith-number/XLL44TIX7ZIGRVEGOVM3B7QDPF/events.json","paper":"https://pith.science/paper/XLL44TIX"},"agent_actions":{"view_html":"https://pith.science/pith/XLL44TIX7ZIGRVEGOVM3B7QDPF","download_json":"https://pith.science/pith/XLL44TIX7ZIGRVEGOVM3B7QDPF.json","view_paper":"https://pith.science/paper/XLL44TIX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.20668&json=true","fetch_graph":"https://pith.science/api/pith-number/XLL44TIX7ZIGRVEGOVM3B7QDPF/graph.json","fetch_events":"https://pith.science/api/pith-number/XLL44TIX7ZIGRVEGOVM3B7QDPF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XLL44TIX7ZIGRVEGOVM3B7QDPF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XLL44TIX7ZIGRVEGOVM3B7QDPF/action/storage_attestation","attest_author":"https://pith.science/pith/XLL44TIX7ZIGRVEGOVM3B7QDPF/action/author_attestation","sign_citation":"https://pith.science/pith/XLL44TIX7ZIGRVEGOVM3B7QDPF/action/citation_signature","submit_replication":"https://pith.science/pith/XLL44TIX7ZIGRVEGOVM3B7QDPF/action/replication_record"}},"created_at":"2026-07-24T00:23:28.514706+00:00","updated_at":"2026-07-24T00:23:28.514706+00:00"}