AgentCompass: A Unified Evaluation Infrastructure for Agent Capabilities
Mirrored from Hugging Face Daily Papers for archival readability. Support the source by reading on the original site.
AgentCompass is an extensible open-source evaluation infrastructure for systematically assessing LLM/VLM agent capabilities.</p>\n","updatedAt":"2026-07-16T08:53:14.689Z","author":{"_id":"659817704dbb962b7c724a61","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/Pe3eKSx-hkkgYAvl_mGx0.png","fullname":"Shufan Jiang","name":"Tsumugii","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.780756950378418},"editors":["Tsumugii"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/Pe3eKSx-hkkgYAvl_mGx0.png"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2607.13705","authors":[{"_id":"6a589a9d69aa0f8878bddc98","name":"Zichen Ding","hidden":false},{"_id":"6a589a9d69aa0f8878bddc99","name":"Jiaye Ge","hidden":false},{"_id":"6a589a9d69aa0f8878bddc9a","name":"Shufan Jiang","hidden":false},{"_id":"6a589a9d69aa0f8878bddc9b","name":"Kai Chen","hidden":false},{"_id":"6a589a9d69aa0f8878bddc9c","name":"Mo Li","hidden":false},{"_id":"6a589a9d69aa0f8878bddc9d","name":"Qingqiu Li","hidden":false},{"_id":"6a589a9d69aa0f8878bddc9e","name":"Zehao Li","hidden":false},{"_id":"6a589a9d69aa0f8878bddc9f","name":"Zonglin Li","hidden":false},{"_id":"6a589a9d69aa0f8878bddca0","name":"Tiaohao Liang","hidden":false},{"_id":"6a589a9d69aa0f8878bddca1","name":"Shudong Liu","hidden":false},{"_id":"6a589a9d69aa0f8878bddca2","name":"Zerun Ma","hidden":false},{"_id":"6a589a9d69aa0f8878bddca3","name":"Zixing Shang","hidden":false},{"_id":"6a589a9d69aa0f8878bddca4","name":"Wenhui Tian","hidden":false},{"_id":"6a589a9d69aa0f8878bddca5","name":"Zun Wang","hidden":false},{"_id":"6a589a9d69aa0f8878bddca6","name":"Liwei Wu","hidden":false},{"_id":"6a589a9d69aa0f8878bddca7","name":"Zhenyu Wu","hidden":false},{"_id":"6a589a9d69aa0f8878bddca8","name":"Jun Xu","hidden":false},{"_id":"6a589a9d69aa0f8878bddca9","name":"Bowen Yang","hidden":false},{"_id":"6a589a9d69aa0f8878bddcaa","name":"Dingbo Yuan","hidden":false},{"_id":"6a589a9d69aa0f8878bddcab","name":"Qi Zhang","hidden":false},{"_id":"6a589a9d69aa0f8878bddcac","name":"Songyang Zhang","hidden":false},{"_id":"6a589a9d69aa0f8878bddcad","name":"Peiheng Zhou","hidden":false},{"_id":"6a589a9d69aa0f8878bddcae","name":"Dongsheng Zhu","hidden":false}],"publishedAt":"2026-07-15T00:00:00.000Z","submittedOnDailyAt":"2026-07-16T00:00:00.000Z","title":"AgentCompass: A Unified Evaluation Infrastructure for Agent Capabilities","submittedOnDailyBy":{"_id":"659817704dbb962b7c724a61","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/Pe3eKSx-hkkgYAvl_mGx0.png","isPro":false,"fullname":"Shufan Jiang","user":"Tsumugii","type":"user","name":"Tsumugii"},"summary":"As Large Language Models (LLMs) evolve into autonomous agents, the need for unified evaluation infrastructure becomes critical. However, current evaluation pipelines remain highly fragmented and tightly coupled, hindering reproducibility and causing redundant engineering. To address this, we introduce AgentCompass, an open-source, lightweight, and extensible infrastructure for evaluating LLM-based agents. AgentCompass organizes the evaluation process around three independent components, namely Benchmark, Harness, and Environment, thereby enabling flexible configurations without requiring the reimplementation of complex execution logic. Furthermore, it features a fault-tolerant asynchronous runtime and comprehensive trajectory analysis tools to transparently diagnose nuanced failure modes like reward-hacking. Natively supporting over 20 benchmarks across five capability dimensions, AgentCompass provides the community with a scalable and reproducible infrastructure for advancing agent research.","upvotes":9,"discussionId":"6a589a9d69aa0f8878bddcaf","githubRepo":"https://github.com/open-compass/agentcompass","githubRepoAddedBy":"user","githubStars":30,"organization":{"_id":"64ef24837b31b8cf66b2e911","name":"opencompass","fullname":"OpenCompass","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/630716d11801ecc7d2595021/OP8C6Z5-G7p6E33wZJm33.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"659817704dbb962b7c724a61","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/Pe3eKSx-hkkgYAvl_mGx0.png","isPro":false,"fullname":"Shufan Jiang","user":"Tsumugii","type":"user"},{"_id":"698309b43ea114b8ee0f921e","avatarUrl":"/avatars/d4c942d84879cfb6663f6ea8f4b16995.svg","isPro":false,"fullname":"Phyo Phyo","user":"ugh-45","type":"user"},{"_id":"620783f24e28382272337ba4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620783f24e28382272337ba4/zkUveQPNiDfYjgGhuFErj.jpeg","isPro":false,"fullname":"GuoLiangTang","user":"Tommy930","type":"user"},{"_id":"65c4eb7cd1dcbd30d86febec","avatarUrl":"/avatars/001c8f02e8ce794b2c21883628b2da72.svg","isPro":false,"fullname":"free-bit","user":"free-bit","type":"user"},{"_id":"66fb03d6b505f1a04c39d935","avatarUrl":"/avatars/e9b830c460ec02037758c9b3469bb8ad.svg","isPro":false,"fullname":"Xuanlang Dai","user":"XuanlangDai","type":"user"},{"_id":"6270324ebecab9e2dcf245de","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6270324ebecab9e2dcf245de/cMbtWSasyNlYc9hvsEEzt.jpeg","isPro":false,"fullname":"Kye Gomez","user":"kye","type":"user"},{"_id":"63ac5701c21e60a3e9b58aa7","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63ac5701c21e60a3e9b58aa7/g6EX7diOpuA94R2ab-rZC.png","isPro":true,"fullname":"Dipankar Sarkar","user":"dipankarsarkar","type":"user"},{"_id":"619f9755da83161f25840698","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/619f9755da83161f25840698/FM421pE1mz5v1YhrxA8ZA.jpeg","isPro":false,"fullname":"Muhammad Umair","user":"umair894","type":"user"},{"_id":"6a2da6c8ca070ee12c6e396c","avatarUrl":"/avatars/0355287dcabaa67dbc7f0b10b87451f9.svg","isPro":false,"fullname":"Joe Mama","user":"JoeMama123123123","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"64ef24837b31b8cf66b2e911","name":"opencompass","fullname":"OpenCompass","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/630716d11801ecc7d2595021/OP8C6Z5-G7p6E33wZJm33.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2607/2607.13705.md","query":{}}">
Papers
arxiv:2607.13705
AgentCompass: A Unified Evaluation Infrastructure for Agent Capabilities
Authors:
,
Abstract
As Large Language Models (LLMs) evolve into autonomous agents, the need for unified evaluation infrastructure becomes critical. However, current evaluation pipelines remain highly fragmented and tightly coupled, hindering reproducibility and causing redundant engineering. To address this, we introduce AgentCompass, an open-source, lightweight, and extensible infrastructure for evaluating LLM-based agents. AgentCompass organizes the evaluation process around three independent components, namely Benchmark, Harness, and Environment, thereby enabling flexible configurations without requiring the reimplementation of complex execution logic. Furthermore, it features a fault-tolerant asynchronous runtime and comprehensive trajectory analysis tools to transparently diagnose nuanced failure modes like reward-hacking. Natively supporting over 20 benchmarks across five capability dimensions, AgentCompass provides the community with a scalable and reproducible infrastructure for advancing agent research.
Community
AgentCompass is an extensible open-source evaluation infrastructure for systematically assessing LLM/VLM agent capabilities.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Get this paper in your agent:
hf papers read 2607.13705 curl -LsSf https://hf.co/cli/install.sh | bash Models citing this paper
No model linking this paper
Cite arxiv.org/abs/2607.13705 in a model README.md to link it from this page.
Datasets citing this paper
No dataset linking this paper
Cite arxiv.org/abs/2607.13705 in a dataset README.md to link it from this page.
Spaces citing this paper
No Space linking this paper
Cite arxiv.org/abs/2607.13705 in a Space README.md to link it from this page.
Collections including this paper
No Collection including this paper
Add this paper to a collection to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.