{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/code/call-llm","entry":"call_llm","source":"Syntology graph, per-sample; not an archive number","read_at":"2026-09-24T18:15:14+00:00","claim":"Names are grouped by exact entry-name string. Same-named routines are NOT asserted to be equivalent; 'ran' means executed on a synthesized fixture, not correctness. n_samples_ran = sum of by_status over every status except 'unverified' (ran_draft_wrong and ran_fixture are failures of Syntology's instrument, not of the code); n_papers_ran = papers with at least one such sample.","status_vocabulary":{"ran_honours":"ran, honoured the contract we drafted","ran_violates":"ran, violated the contract we drafted","ran_draft_wrong":"ran; our contract draft was wrong, not the code","ran_fixture":"ran; our fixture could not drive it","ran":"ran on a synthesized input","unverified":"unverified (harvested, no recorded run)"},"n_papers":14,"n_papers_ran":7,"units":"n_samples, n_samples_ran, n_samples_fingerprinted and by_status count distinct code bodies (code_sha256); n_places and n_places_pointer_only count places, one per (paper, code body) pair, which is also the unit of the samples list","n_samples":15,"n_samples_ran":6,"n_samples_fingerprinted":0,"n_places":16,"n_places_pointer_only":6,"by_status":{"ran_honours":0,"ran_violates":0,"ran_draft_wrong":3,"ran_fixture":0,"ran":3,"unverified":9},"syntology":{"atlas_url":null,"mcp":null,"mcp_per_sample":{"tool":"get_code","arguments_in":"samples[].mcp_get_code"},"developers":"https://syntology.ai/developers"},"samples":[{"arxiv_id":"2609.00621","paper":"/paper/arxiv-2609-00621","title":"Control-Data Flow Separation: Stable Prompt Optimization in Multi-Agent LLMs","date":null,"month_inferred_from_arxiv_id":"2026-09","title_source":"syntology","repo":"yuntian-group/cdsep","path":"cdsep/providers.py","file_url":"https://github.com/yuntian-group/cdsep/blob/HEAD/cdsep/providers.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"809e9f49f28a6edb","mcp_get_code":{"code_sha256":"809e9f49f28a6edb"}},{"arxiv_id":"2608.25973","paper":"/paper/arxiv-2608-25973","title":"SciMIF: Understanding Multimodal Instruction Following in Scientific Domains","date":null,"month_inferred_from_arxiv_id":"2026-08","title_source":"syntology","repo":"shenye7436/SciMIF","path":"data_construction.py","file_url":"https://github.com/shenye7436/SciMIF/blob/HEAD/data_construction.py","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"35ba0bc7a58f7218","mcp_get_code":{"code_sha256":"35ba0bc7a58f7218"}},{"arxiv_id":"2608.08722","paper":"/paper/arxiv-2608-08722","title":"Gaming Without an Attacker: Benchmark Fingerprinting in LLM-Driven Search Under Selection Pressure","date":null,"month_inferred_from_arxiv_id":"2026-08","title_source":"syntology","repo":"vicgalle/kernel-fingerprinting","path":"metal_zk/llm.py","file_url":"https://github.com/vicgalle/kernel-fingerprinting/blob/HEAD/metal_zk/llm.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"f740470510a7fa9e","mcp_get_code":{"code_sha256":"f740470510a7fa9e"}},{"arxiv_id":"2608.06477","paper":"/paper/arxiv-2608-06477","title":"StepJack: Benchmarking Computer-Use Agent Safety Against Multi-Step Indirect Prompt Injection","date":null,"month_inferred_from_arxiv_id":"2026-08","title_source":"syntology","repo":"BorealisAI/StepJack","path":"auto_decomp/iterative_refinement.py","file_url":"https://github.com/BorealisAI/StepJack/blob/HEAD/auto_decomp/iterative_refinement.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"GPL-3.0","inline_ok":false,"code_sha256_prefix":"db78ed8a3233ab84","mcp_get_code":{"code_sha256":"db78ed8a3233ab84"}},{"arxiv_id":"2606.31191","paper":"/paper/arxiv-2606-31191","title":"ISM: Self-Improving Strategy Memory for Continual Mathematical Reasoning","date":null,"month_inferred_from_arxiv_id":"2026-06","title_source":"syntology","repo":"pdx97/ISM","path":"feature_extractor_llm.py","file_url":"https://github.com/pdx97/ISM/blob/HEAD/feature_extractor_llm.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"eb146e366a08a7c5","mcp_get_code":{"code_sha256":"eb146e366a08a7c5"}},{"arxiv_id":"2606.26327","paper":"/paper/arxiv-2606-26327","title":"EVOM: Agentic Meta-Evolution of Actor-Critic Architectures for Reinforcement Learning","date":null,"month_inferred_from_arxiv_id":"2026-06","title_source":"syntology","repo":"xiaofangxd/EVOM","path":"evom.py","file_url":"https://github.com/xiaofangxd/EVOM/blob/HEAD/evom.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"82e34fa3d741ddb4","mcp_get_code":{"code_sha256":"82e34fa3d741ddb4"}},{"arxiv_id":"2605.14665","paper":"/paper/arxiv-2605-14665","title":"Falkor-IRAC: Graph-Constrained Generation for Verified Legal Reasoning in Indian Judicial AI","date":null,"month_inferred_from_arxiv_id":"2026-05","title_source":"syntology","repo":"joyboseroy/falkor-irac","path":"ingestion/irac_parser.py","file_url":"https://github.com/joyboseroy/falkor-irac/blob/HEAD/ingestion/irac_parser.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"c222c3061de78f7f","mcp_get_code":{"code_sha256":"c222c3061de78f7f"}},{"arxiv_id":"2605.14665","paper":"/paper/arxiv-2605-14665","title":"Falkor-IRAC: Graph-Constrained Generation for Verified Legal Reasoning in Indian Judicial AI","date":null,"month_inferred_from_arxiv_id":"2026-05","title_source":"syntology","repo":"joyboseroy/falkor-irac","path":"agents/orchestrator.py","file_url":"https://github.com/joyboseroy/falkor-irac/blob/HEAD/agents/orchestrator.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"46783c92264f8631","mcp_get_code":{"code_sha256":"46783c92264f8631"}},{"arxiv_id":"2605.09708","paper":"/paper/arxiv-2605-09708","title":"Metal-Sci: A Scientific Compute Benchmark for Evolutionary LLM Kernel Search on Apple Silicon","date":null,"month_inferred_from_arxiv_id":"2026-05","title_source":"syntology","repo":"vicgalle/metal-sci-kernels","path":"metal_kernels/llm.py","file_url":"https://github.com/vicgalle/metal-sci-kernels/blob/HEAD/metal_kernels/llm.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"f740470510a7fa9e","mcp_get_code":{"code_sha256":"f740470510a7fa9e"}},{"arxiv_id":"2604.20930","paper":"/paper/arxiv-2604-20930","title":"SafeRedirect: Defeating Internal Safety Collapse via Task-Completion Redirection in Frontier LLMs","date":null,"month_inferred_from_arxiv_id":"2026-04","title_source":"syntology","repo":"fzjcdt/SafeRedirect","path":"run_safe_redirect.py","file_url":"https://github.com/fzjcdt/SafeRedirect/blob/HEAD/run_safe_redirect.py","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"c01d2fbac2241f4f","mcp_get_code":{"code_sha256":"c01d2fbac2241f4f"}},{"arxiv_id":"2603.04855","paper":"/paper/arxiv-2603-04855","title":"HACHIMI: Scalable and Controllable Student Persona Generation via Orchestrated Agents","date":null,"month_inferred_from_arxiv_id":"2026-03","title_source":"syntology","repo":"ZeroLoss-Lab/HACHIMI","path":"app_for_GPT4.1.py","file_url":"https://github.com/ZeroLoss-Lab/HACHIMI/blob/HEAD/app_for_GPT4.1.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"abf2fd85dc0391da","mcp_get_code":{"code_sha256":"abf2fd85dc0391da"}},{"arxiv_id":"2602.10715","paper":"/paper/arxiv-2602-10715","title":"Locomo-Plus: Beyond-Factual Cognitive Memory Evaluation Framework for LLM Agents","date":null,"month_inferred_from_arxiv_id":"2026-02","title_source":"syntology","repo":"xjtuleeyf/Locomo-Plus","path":"evaluation_framework/task_eval/llm_as_judge.py","file_url":"https://github.com/xjtuleeyf/Locomo-Plus/blob/HEAD/evaluation_framework/task_eval/llm_as_judge.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"5c5765a0c9007c76","mcp_get_code":{"code_sha256":"5c5765a0c9007c76"}},{"arxiv_id":"2507.02592","paper":"/paper/websailor-navigating-super-human-reasoning","title":"WebSailor: Navigating Super-human Reasoning for Web Agent","date":"2025-07-03","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"alibaba-nlp/webagent","path":"WebAgent/NestBrowse/utils.py","file_url":"https://github.com/alibaba-nlp/webagent/blob/HEAD/WebAgent/NestBrowse/utils.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"code_sha256_prefix":"cc601f60e63e7a5c","mcp_get_code":{"code_sha256":"cc601f60e63e7a5c"}},{"arxiv_id":"2507.02592","paper":"/paper/websailor-navigating-super-human-reasoning","title":"WebSailor: Navigating Super-human Reasoning for Web Agent","date":"2025-07-03","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"alibaba-nlp/webagent","path":"WebAgent/ParallelMuse/functionality_specified_partial_rollout.py","file_url":"https://github.com/alibaba-nlp/webagent/blob/HEAD/WebAgent/ParallelMuse/functionality_specified_partial_rollout.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"code_sha256_prefix":"ffcb6da25872786e","mcp_get_code":{"code_sha256":"ffcb6da25872786e"}},{"arxiv_id":"2504.13079","paper":"/paper/retrieval-augmented-generation-with-3","title":"Retrieval-Augmented Generation with Conflicting Evidence","date":"2025-04-17","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"hannight/ramdocs","path":"run_madam_rag.py","file_url":"https://github.com/hannight/ramdocs/blob/HEAD/run_madam_rag.py","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"2d6010580b048050","mcp_get_code":{"code_sha256":"2d6010580b048050"}},{"arxiv_id":"2304.07590","paper":"/paper/self-collaboration-code-generation-via","title":"Self-collaboration Code Generation via ChatGPT","date":null,"month_inferred_from_arxiv_id":"2023-04","title_source":"archive","repo":"YihongDong/Self-collaboration-Code-Generation","path":"core/backend.py","file_url":"https://github.com/YihongDong/Self-collaboration-Code-Generation/blob/HEAD/core/backend.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"6b210bdeeae33ffc","mcp_get_code":{"code_sha256":"6b210bdeeae33ffc"}}]}