{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:O3I442MWLWJANGOYIWLTN5FSAP","short_pith_number":"pith:O3I442MW","schema_version":"1.0","canonical_sha256":"76d1ce69965d920699d8459736f4b203e4a0ec2738820c5f561b0c7fad357ff8","source":{"kind":"arxiv","id":"2509.05755","version":6},"attestation_state":"computed","paper":{"title":"Red-Teaming Coding Agents from a Tool-Invocation Perspective: An Empirical Security Assessment","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Dongdong She, Kaikai Zhang, Mingyu Luo, Ping Chen, Shuai Wang, Yuchong Xie, Yu Liu, Zesen Liu, Zhixiang Zhang, Zongjie Li","submitted_at":"2025-09-06T15:48:49Z","abstract_excerpt":"Coding agents powered by large language models are becoming central modules of modern IDEs, helping users perform complex tasks by invoking tools. While powerful, tool invocation opens a substantial attack surface. Prior work has demonstrated attacks against general-purpose and domain-specific agents, but none have focused on the security risks of tool invocation in coding agents. To fill this gap, we conduct the first systematic red-teaming of six popular real-world coding agents: Cursor, Claude Code, Copilot, Windsurf, Cline, and Trae. Our red-teaming proceeds in two phases. In Phase 1, we p"},"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":"2509.05755","kind":"arxiv","version":6},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-09-06T15:48:49Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c0b7475b20a88bd87eb25a94eb2984c04704ea8689125b8c8cd82bc5a4434995","abstract_canon_sha256":"3ccc9f2c0aaf40010e35507ddf30725a3d6786e00f2707e0d091c988cacf9a22"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-30T01:17:25.836352Z","signature_b64":"jrIn0BepnKdiQHzhBQkVDnemoDbHlFY02OFiFkSDsRmK1W9uBQGfza0Yi3ve926CbaUGHnROF4/McHexibvlCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"76d1ce69965d920699d8459736f4b203e4a0ec2738820c5f561b0c7fad357ff8","last_reissued_at":"2026-06-30T01:17:25.835665Z","signature_status":"signed_v1","first_computed_at":"2026-06-30T01:17:25.835665Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Red-Teaming Coding Agents from a Tool-Invocation Perspective: An Empirical Security Assessment","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Dongdong She, Kaikai Zhang, Mingyu Luo, Ping Chen, Shuai Wang, Yuchong Xie, Yu Liu, Zesen Liu, Zhixiang Zhang, Zongjie Li","submitted_at":"2025-09-06T15:48:49Z","abstract_excerpt":"Coding agents powered by large language models are becoming central modules of modern IDEs, helping users perform complex tasks by invoking tools. While powerful, tool invocation opens a substantial attack surface. Prior work has demonstrated attacks against general-purpose and domain-specific agents, but none have focused on the security risks of tool invocation in coding agents. To fill this gap, we conduct the first systematic red-teaming of six popular real-world coding agents: Cursor, Claude Code, Copilot, Windsurf, Cline, and Trae. Our red-teaming proceeds in two phases. In Phase 1, we p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.05755","kind":"arxiv","version":6},"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/2509.05755/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":"2509.05755","created_at":"2026-06-30T01:17:25.835748+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.05755v6","created_at":"2026-06-30T01:17:25.835748+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.05755","created_at":"2026-06-30T01:17:25.835748+00:00"},{"alias_kind":"pith_short_12","alias_value":"O3I442MWLWJA","created_at":"2026-06-30T01:17:25.835748+00:00"},{"alias_kind":"pith_short_16","alias_value":"O3I442MWLWJANGOY","created_at":"2026-06-30T01:17:25.835748+00:00"},{"alias_kind":"pith_short_8","alias_value":"O3I442MW","created_at":"2026-06-30T01:17:25.835748+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":5,"sample":[{"citing_arxiv_id":"2605.24069","citing_title":"When the Manual Lies: A Realistic Benchmark to Evaluate MCP Poisoning Attacks for LLM Agents","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2606.10749","citing_title":"Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluation","ref_index":208,"is_internal_anchor":true},{"citing_arxiv_id":"2509.06572","citing_title":"Parasites in the Toolchain: A Large-Scale Analysis of Attacks on the MCP Ecosystem","ref_index":50,"is_internal_anchor":true},{"citing_arxiv_id":"2603.09002","citing_title":"Security Considerations for Multi-agent Systems","ref_index":279,"is_internal_anchor":true},{"citing_arxiv_id":"2605.02187","citing_title":"When Alignment Isn't Enough: Response-Path Attacks on LLM Agents","ref_index":4,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/O3I442MWLWJANGOYIWLTN5FSAP","json":"https://pith.science/pith/O3I442MWLWJANGOYIWLTN5FSAP.json","graph_json":"https://pith.science/api/pith-number/O3I442MWLWJANGOYIWLTN5FSAP/graph.json","events_json":"https://pith.science/api/pith-number/O3I442MWLWJANGOYIWLTN5FSAP/events.json","paper":"https://pith.science/paper/O3I442MW"},"agent_actions":{"view_html":"https://pith.science/pith/O3I442MWLWJANGOYIWLTN5FSAP","download_json":"https://pith.science/pith/O3I442MWLWJANGOYIWLTN5FSAP.json","view_paper":"https://pith.science/paper/O3I442MW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.05755&json=true","fetch_graph":"https://pith.science/api/pith-number/O3I442MWLWJANGOYIWLTN5FSAP/graph.json","fetch_events":"https://pith.science/api/pith-number/O3I442MWLWJANGOYIWLTN5FSAP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/O3I442MWLWJANGOYIWLTN5FSAP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/O3I442MWLWJANGOYIWLTN5FSAP/action/storage_attestation","attest_author":"https://pith.science/pith/O3I442MWLWJANGOYIWLTN5FSAP/action/author_attestation","sign_citation":"https://pith.science/pith/O3I442MWLWJANGOYIWLTN5FSAP/action/citation_signature","submit_replication":"https://pith.science/pith/O3I442MWLWJANGOYIWLTN5FSAP/action/replication_record"}},"created_at":"2026-06-30T01:17:25.835748+00:00","updated_at":"2026-06-30T01:17:25.835748+00:00"}