{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:4TSBJ2Q322ZW2H3KGJOUWRE4NJ","short_pith_number":"pith:4TSBJ2Q3","schema_version":"1.0","canonical_sha256":"e4e414ea1bd6b36d1f6a325d4b449c6a7a245b5ea65f879901aba54b7a286b7c","source":{"kind":"arxiv","id":"2508.20453","version":1},"attestation_state":"computed","paper":{"title":"MCP-Bench: Benchmarking Tool-Using LLM Agents with Complex Real-World Tasks via MCP Servers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alireza Rezazadeh, Ankit Shah, Aolin Ding, Cheng-En Wu, Eugene Siow, Hemani Patel, Qi Chang, Quan Liu, Shashank Biju, Yujia Bao, Zhenting Wang","submitted_at":"2025-08-28T05:58:57Z","abstract_excerpt":"We introduce MCP-Bench, a benchmark for evaluating large language models (LLMs) on realistic, multi-step tasks that demand tool use, cross-tool coordination, precise parameter control, and planning/reasoning for solving tasks. Built on the Model Context Protocol (MCP), MCP-Bench connects LLMs to 28 representative live MCP servers spanning 250 tools across domains such as finance, traveling, scientific computing, and academic search. Unlike prior API-based benchmarks, each MCP server provides a set of complementary tools designed to work together, enabling the construction of authentic, multi-s"},"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":"2508.20453","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-08-28T05:58:57Z","cross_cats_sorted":[],"title_canon_sha256":"8aa32defad70b5c70bde0b758850099a7318c85a56ace40b75f28d33d698a3a2","abstract_canon_sha256":"48df1b627fb066d163ab6a02caebad220f9c60c86112b7974149aa4ef3aedc3d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:00:36.858643Z","signature_b64":"2v+vMtcraXGs4yNHM8TZqqcg40yDOuVpG5p68HoVDTYI0RaBHn/36PN6OwRsaDP6CoCrO4odbdyX0cGffz8pBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e4e414ea1bd6b36d1f6a325d4b449c6a7a245b5ea65f879901aba54b7a286b7c","last_reissued_at":"2026-07-05T12:00:36.858087Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:00:36.858087Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MCP-Bench: Benchmarking Tool-Using LLM Agents with Complex Real-World Tasks via MCP Servers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alireza Rezazadeh, Ankit Shah, Aolin Ding, Cheng-En Wu, Eugene Siow, Hemani Patel, Qi Chang, Quan Liu, Shashank Biju, Yujia Bao, Zhenting Wang","submitted_at":"2025-08-28T05:58:57Z","abstract_excerpt":"We introduce MCP-Bench, a benchmark for evaluating large language models (LLMs) on realistic, multi-step tasks that demand tool use, cross-tool coordination, precise parameter control, and planning/reasoning for solving tasks. Built on the Model Context Protocol (MCP), MCP-Bench connects LLMs to 28 representative live MCP servers spanning 250 tools across domains such as finance, traveling, scientific computing, and academic search. Unlike prior API-based benchmarks, each MCP server provides a set of complementary tools designed to work together, enabling the construction of authentic, multi-s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.20453","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/2508.20453/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":"2508.20453","created_at":"2026-07-05T12:00:36.858149+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.20453v1","created_at":"2026-07-05T12:00:36.858149+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.20453","created_at":"2026-07-05T12:00:36.858149+00:00"},{"alias_kind":"pith_short_12","alias_value":"4TSBJ2Q322ZW","created_at":"2026-07-05T12:00:36.858149+00:00"},{"alias_kind":"pith_short_16","alias_value":"4TSBJ2Q322ZW2H3K","created_at":"2026-07-05T12:00:36.858149+00:00"},{"alias_kind":"pith_short_8","alias_value":"4TSBJ2Q3","created_at":"2026-07-05T12:00:36.858149+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":30,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22385","citing_title":"MetaPS: Adaptive Programmatic Strategy Selection for Market Agents","ref_index":99,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17819","citing_title":"A Framework for Evaluating Agentic Skills at Scale","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2606.16591","citing_title":"SING: Synthetic Intention Graph for Scalable Active Tool Discovery in LLM Agents","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12908","citing_title":"SENTINEL: Failure-Driven Reinforcement Learning for Training Tool-Using Language Model Agents","ref_index":54,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09182","citing_title":"Understanding How Enterprises Adopt the Model Context Protocol for LLM-Driven Software Engineering","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2606.10209","citing_title":"Less Context, Better Agents: Efficient Context Engineering for Long-Horizon Tool-Using LLM Agents","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.31308","citing_title":"TraceGraph: Shared Decision Landscapes for Diagnosing and Improving Agent Trajectories","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30785","citing_title":"Learning Agent-Compatible Context Management for Long-Horizon Tasks","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12191","citing_title":"Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application","ref_index":136,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12674","citing_title":"Evoflux: Inference-Time Evolution of Executable Tool Workflows for Compact Agents","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2602.00933","citing_title":"MCP-Atlas: A Large-Scale Benchmark for Tool-Use Competency with Real MCP Servers","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15104","citing_title":"From Text to Voice: A Reproducible and Verifiable Framework for Evaluating Tool Calling LLM Agents","ref_index":270,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10787","citing_title":"ComplexMCP: Evaluation of LLM Agents in Dynamic, Interdependent, and Large-Scale Tool Sandbox","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07926","citing_title":"AgentEscapeBench: Evaluating Out-of-Domain Tool-Grounded Reasoning in LLM Agents","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16909","citing_title":"TOBench: A Task-Oriented Omni-Modal Benchmark for Real-World Tool-Using Agents","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2602.00933","citing_title":"MCP-Atlas: A Large-Scale Benchmark for Tool-Use Competency with Real MCP Servers","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2602.11224","citing_title":"Agent-Diff: Benchmarking LLM Agents on Enterprise API Tasks via Code Execution with State-Diff-Based Evaluation","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13064","citing_title":"Red Skills or Blue Skills? A Dive Into Skills Published on ClawHub","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2604.01532","citing_title":"PHMForge: Evaluating LLM Agents on Industrial Prognostics through MCP-Native, Algorithm-Grounded Tools","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2604.28139","citing_title":"Claw-Eval-Live: A Live Agent Benchmark for Evolving Real-World Workflows","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10912","citing_title":"WildClawBench: A Benchmark for Real-World, Long-Horizon Agent Evaluation","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10787","citing_title":"ComplexMCP: Evaluation of LLM Agents in Dynamic, Interdependent, and Large-Scale Tool Sandbox","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23781","citing_title":"ClawMark: A Living-World Benchmark for Multi-Turn, Multi-Day, Multimodal Coworker Agents","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02240","citing_title":"PhysicianBench: Evaluating LLM Agents in Real-World EHR Environments","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19657","citing_title":"An AI Agent Execution Environment to Safeguard User Data","ref_index":74,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4TSBJ2Q322ZW2H3KGJOUWRE4NJ","json":"https://pith.science/pith/4TSBJ2Q322ZW2H3KGJOUWRE4NJ.json","graph_json":"https://pith.science/api/pith-number/4TSBJ2Q322ZW2H3KGJOUWRE4NJ/graph.json","events_json":"https://pith.science/api/pith-number/4TSBJ2Q322ZW2H3KGJOUWRE4NJ/events.json","paper":"https://pith.science/paper/4TSBJ2Q3"},"agent_actions":{"view_html":"https://pith.science/pith/4TSBJ2Q322ZW2H3KGJOUWRE4NJ","download_json":"https://pith.science/pith/4TSBJ2Q322ZW2H3KGJOUWRE4NJ.json","view_paper":"https://pith.science/paper/4TSBJ2Q3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.20453&json=true","fetch_graph":"https://pith.science/api/pith-number/4TSBJ2Q322ZW2H3KGJOUWRE4NJ/graph.json","fetch_events":"https://pith.science/api/pith-number/4TSBJ2Q322ZW2H3KGJOUWRE4NJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4TSBJ2Q322ZW2H3KGJOUWRE4NJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4TSBJ2Q322ZW2H3KGJOUWRE4NJ/action/storage_attestation","attest_author":"https://pith.science/pith/4TSBJ2Q322ZW2H3KGJOUWRE4NJ/action/author_attestation","sign_citation":"https://pith.science/pith/4TSBJ2Q322ZW2H3KGJOUWRE4NJ/action/citation_signature","submit_replication":"https://pith.science/pith/4TSBJ2Q322ZW2H3KGJOUWRE4NJ/action/replication_record"}},"created_at":"2026-07-05T12:00:36.858149+00:00","updated_at":"2026-07-05T12:00:36.858149+00:00"}