{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:XNZKFRXSYVVT2ZHHGF7HSCTEAN","short_pith_number":"pith:XNZKFRXS","canonical_record":{"source":{"id":"2606.17838","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-16T12:06:27Z","cross_cats_sorted":[],"title_canon_sha256":"a3e33bc970ce83bd6d308266fe08832ff89244801859c0f62717bab474ae719d","abstract_canon_sha256":"d413e7f01c3241e94c182fa0cf916c2dfa0d69ce159daa2acc453eae2d865784"},"schema_version":"1.0"},"canonical_sha256":"bb72a2c6f2c56b3d64e7317e790a640342a2d632b5ef4ff9742c4546e0ca6907","source":{"kind":"arxiv","id":"2606.17838","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.17838","created_at":"2026-06-19T16:10:41Z"},{"alias_kind":"arxiv_version","alias_value":"2606.17838v1","created_at":"2026-06-19T16:10:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.17838","created_at":"2026-06-19T16:10:41Z"},{"alias_kind":"pith_short_12","alias_value":"XNZKFRXSYVVT","created_at":"2026-06-19T16:10:41Z"},{"alias_kind":"pith_short_16","alias_value":"XNZKFRXSYVVT2ZHH","created_at":"2026-06-19T16:10:41Z"},{"alias_kind":"pith_short_8","alias_value":"XNZKFRXS","created_at":"2026-06-19T16:10:41Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:XNZKFRXSYVVT2ZHHGF7HSCTEAN","target":"record","payload":{"canonical_record":{"source":{"id":"2606.17838","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-16T12:06:27Z","cross_cats_sorted":[],"title_canon_sha256":"a3e33bc970ce83bd6d308266fe08832ff89244801859c0f62717bab474ae719d","abstract_canon_sha256":"d413e7f01c3241e94c182fa0cf916c2dfa0d69ce159daa2acc453eae2d865784"},"schema_version":"1.0"},"canonical_sha256":"bb72a2c6f2c56b3d64e7317e790a640342a2d632b5ef4ff9742c4546e0ca6907","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:10:41.025669Z","signature_b64":"Z95xU3y+l9Bh/LO8hkgPePIkfE5kmp0OelEP54fOV+DUE9Die/b6oLE6Fodb75Pe9mT86HI7KxZ+/Tk1ky5BBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bb72a2c6f2c56b3d64e7317e790a640342a2d632b5ef4ff9742c4546e0ca6907","last_reissued_at":"2026-06-19T16:10:41.025243Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:10:41.025243Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2606.17838","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-19T16:10:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"s8H+XeVmUIKLy6/OjlAQVUvQnIvIYmJT+HOyT9K8Wr1xZWs+Mh+Gn4TDom7LjHMVeOpVHV8cK4PJxNKd9PHPDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-21T13:00:11.209925Z"},"content_sha256":"968bf7b72a380a70ca2c06c3730fe1821b5bdd4bfe82fbc5148d6f589f26a853","schema_version":"1.0","event_id":"sha256:968bf7b72a380a70ca2c06c3730fe1821b5bdd4bfe82fbc5148d6f589f26a853"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:XNZKFRXSYVVT2ZHHGF7HSCTEAN","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Environment-Grounded Automated Prompt Optimization for LLM Game Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Lukas Fehring, Marius Lindauer, Matthias Feurer, Rean Clive Fernandes, Theresa Eimer","submitted_at":"2026-06-16T12:06:27Z","abstract_excerpt":"LLM agents in interactive environments are highly sensitive to their prompts, yet prompt engineering remains a manual, task-specific process. We introduce an automated prompt optimization framework for LLM agents that decomposes the observation-to-action pipeline into a goal-conditioned descriptor agent and an action selection agent, and iteratively refines each module's prompt through an LLM-driven evolutionary loop guided by environment returns. We propose a behavior analyzer to attribute episode outcomes to specific prompt components, and a mutator to propose targeted revisions to the promp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.17838","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/2606.17838/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-19T16:10:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"NqOQhlj/UYtLPq5rq+7Skvv7sU0Fk4hOaUmV6S+QdLUo/bOemK27S+13FZRnTfy7b8nu/pDPoR8y3/5TqsTMBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-21T13:00:11.210289Z"},"content_sha256":"a5e2a26becc529563ee55001661bac1984718fc7a474ce870f13530623d972b1","schema_version":"1.0","event_id":"sha256:a5e2a26becc529563ee55001661bac1984718fc7a474ce870f13530623d972b1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/XNZKFRXSYVVT2ZHHGF7HSCTEAN/bundle.json","state_url":"https://pith.science/pith/XNZKFRXSYVVT2ZHHGF7HSCTEAN/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/XNZKFRXSYVVT2ZHHGF7HSCTEAN/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-21T13:00:11Z","links":{"resolver":"https://pith.science/pith/XNZKFRXSYVVT2ZHHGF7HSCTEAN","bundle":"https://pith.science/pith/XNZKFRXSYVVT2ZHHGF7HSCTEAN/bundle.json","state":"https://pith.science/pith/XNZKFRXSYVVT2ZHHGF7HSCTEAN/state.json","well_known_bundle":"https://pith.science/.well-known/pith/XNZKFRXSYVVT2ZHHGF7HSCTEAN/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:XNZKFRXSYVVT2ZHHGF7HSCTEAN","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"d413e7f01c3241e94c182fa0cf916c2dfa0d69ce159daa2acc453eae2d865784","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-16T12:06:27Z","title_canon_sha256":"a3e33bc970ce83bd6d308266fe08832ff89244801859c0f62717bab474ae719d"},"schema_version":"1.0","source":{"id":"2606.17838","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.17838","created_at":"2026-06-19T16:10:41Z"},{"alias_kind":"arxiv_version","alias_value":"2606.17838v1","created_at":"2026-06-19T16:10:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.17838","created_at":"2026-06-19T16:10:41Z"},{"alias_kind":"pith_short_12","alias_value":"XNZKFRXSYVVT","created_at":"2026-06-19T16:10:41Z"},{"alias_kind":"pith_short_16","alias_value":"XNZKFRXSYVVT2ZHH","created_at":"2026-06-19T16:10:41Z"},{"alias_kind":"pith_short_8","alias_value":"XNZKFRXS","created_at":"2026-06-19T16:10:41Z"}],"graph_snapshots":[{"event_id":"sha256:a5e2a26becc529563ee55001661bac1984718fc7a474ce870f13530623d972b1","target":"graph","created_at":"2026-06-19T16:10:41Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2606.17838/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"LLM agents in interactive environments are highly sensitive to their prompts, yet prompt engineering remains a manual, task-specific process. We introduce an automated prompt optimization framework for LLM agents that decomposes the observation-to-action pipeline into a goal-conditioned descriptor agent and an action selection agent, and iteratively refines each module's prompt through an LLM-driven evolutionary loop guided by environment returns. We propose a behavior analyzer to attribute episode outcomes to specific prompt components, and a mutator to propose targeted revisions to the promp","authors_text":"Lukas Fehring, Marius Lindauer, Matthias Feurer, Rean Clive Fernandes, Theresa Eimer","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-16T12:06:27Z","title":"Environment-Grounded Automated Prompt Optimization for LLM Game Agents"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.17838","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:968bf7b72a380a70ca2c06c3730fe1821b5bdd4bfe82fbc5148d6f589f26a853","target":"record","created_at":"2026-06-19T16:10:41Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"d413e7f01c3241e94c182fa0cf916c2dfa0d69ce159daa2acc453eae2d865784","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-16T12:06:27Z","title_canon_sha256":"a3e33bc970ce83bd6d308266fe08832ff89244801859c0f62717bab474ae719d"},"schema_version":"1.0","source":{"id":"2606.17838","kind":"arxiv","version":1}},"canonical_sha256":"bb72a2c6f2c56b3d64e7317e790a640342a2d632b5ef4ff9742c4546e0ca6907","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"bb72a2c6f2c56b3d64e7317e790a640342a2d632b5ef4ff9742c4546e0ca6907","first_computed_at":"2026-06-19T16:10:41.025243Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-19T16:10:41.025243Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Z95xU3y+l9Bh/LO8hkgPePIkfE5kmp0OelEP54fOV+DUE9Die/b6oLE6Fodb75Pe9mT86HI7KxZ+/Tk1ky5BBA==","signature_status":"signed_v1","signed_at":"2026-06-19T16:10:41.025669Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.17838","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:968bf7b72a380a70ca2c06c3730fe1821b5bdd4bfe82fbc5148d6f589f26a853","sha256:a5e2a26becc529563ee55001661bac1984718fc7a474ce870f13530623d972b1"],"state_sha256":"97970d58813a6e6d56b9eb038defe40a204207ba47eb7261d6b777cb4b343b1d"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"w9Iss93pw6kMe9Xo8/IkVncWIR0AuNz7xPNh0KGzIUZjvQ7yG/yvMFyncRJZXRInw56RqOMwUyYCcgYVzurSBQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-21T13:00:11.212258Z","bundle_sha256":"3a01369e41ae5175d387368c2eea10b0c0447a9375cf237c21eedcc5f22127dd"}}