AgentAudit is an open, extensible framework that evaluates the full lifecycle of AI agents, assessing planning, tool selection, execution, memory, and reasoning across ten dimensions such as instruction integrity, security, and alignment. Unlike existing benchmarks that focus on single aspects, AgentAudit analyzes the entire execution trace to attribute failures to specific stages. The framework was tested on five large language models, revealing significant differences in trustworthiness even among models with similar task‑completion performance.
By Shrey Nag, Sachita, Abhishek Kumar Singh, Lipi Goel, Rajeshwar Singh Janwar
arXiv:2609.21841v1 Announce Type: new
Abstract: Frontier language models now produce professional deliverables that expert graders judge to match human work on a substantial share of economically val...
By Abbas Raza Ali, Muhammad Ajmal Siddiqui, Moona Zahid
arXiv:2608. 00794v2 Announce Type: replace Abstract: Agentic AI evaluation pipelines produce benchmark scores that justify deployment decisions, safety certifications, and regulatory compliance claims.
By William Caban
The article discusses how AI agents are transforming software development by autonomously creating pull requests, selecting dependencies, refactoring code, interpreting test failures, and managing infrastructure across the SDLC. It highlights the lack of accountability for these transient processes, noting that traditional tools like git logs and governance frameworks assume human decision‑making, which no longer applies. To address this, the authors propose Agentic Process Evidence (APE), a signed, tamper‑evident, machine‑readable attestation that links AI agent sessions to their outcomes, enabling policy‑as‑code controls and human oversight only when necessary.
By Yuval Kainan, Carmit Hershman
arXiv:2607. 16345v1 Announce Type: cross Abstract: Modern agentic systems increasingly rely on skills: installable packages of natural language and code that teach an LLM agent to perform a domain task.
By Tejas Singh Anand, Yuet Ying Christina Wang, Wanting Jiang, Steve Masson, Tian Zheng, Bingjie Zhou
EIO-Agents introduces an open specification for interoperable AI agent evaluation, consisting of two layers: the Evaluation Intelligence Ontology (EIO) and the Portable Evaluation Record (PER). The EIO layer supplies a semantic framework that defines typed evidence, behavioral predicates, evidence contracts, claims, witness rules, proof status, recurrence, and computable derivations for metrics, findings, controls, and decisions such as PASS, REVIEW, or BLOCK. The PER layer offers a canonical, content‑addressed record of an evaluation that preserves the evidence‑to‑decision chain, enabling re‑derivation, explanation, and verification of the evaluation outcome.
By Fouad Bousetouane
The paper introduces TraceTrial-Bench, a benchmark that evaluates the reliability of AI agent harnesses across seven components—context, execution, governance, lifecycle, observability, tools, and verification—using 95 expert-level tasks with executable or state-based ground truth. It employs a Trajectory Tracing module to record agent execution, a two-layer evaluation framework to separately assess agent performance and harness reliability, and a Multi-source Evidence Acquisition module to gather evidence for LLM-based evaluation. The benchmark’s analysis of 3,420 runs across four harnesses and nine models shows harness scores between 59.8% and 72.0%, with 31.1% of runs succeeding despite failing at least one harness criterion, highlighting the need for auditable and reproducible harness evaluation.
By Zhiyong Cao, Hengjie Zhu, Yongxin Wang, Xuchen Li, Jintao Li, Shuxia Liu, Dunqiang Liu, Bo Wang, Kairan Zhang, Cong Lin, JiangDi Yu, Siyuan Liu, Sijia Yao, Yi-Liang Wei, Lingxu Ma, Aijia Wang, Anxin Chen, Zixian Wei, Xiang Ruan, Ruizhi Xu, Ruqian Shi, Baihua Xiao, Sicheng Zhou
The paper proposes that self‑evolving large‑language‑model agents can only be safely reviewed if their changes are confined to the runtime harness—instruction text, tool‑call logic, and primitive composition—while the model weights remain fixed. A dual‑loop engine with an admission gate records a hash‑chained diff before deployment, and simulation experiments show that this gate admits only safe changes, preventing error increases and maintaining false‑positive rates. When the gate is removed, many harmful changes slip through and missed flags rise, illustrating the gate’s protective role.
By Ravil Akhtyamov
arXiv:2608. 11274v1 Announce Type: cross Abstract: The dominant paradigm treats AI safety as a property to be instilled during model training via RLHF, DPO, or Constitutional AI.
By Albus W. Ng, Yi Han, Jusheng Zhang, Wenhao Wang
arXiv:2606. 00448v1 Announce Type: cross Abstract: LLM agents increasingly rely on community-contributed skills that expand an agent's operational capability set.
By Su Wang, Pin Qian, Yihang Chen, Junxian You, Xiaoyuan Wang, Xiaochong Jiang, Lifei Liu, Haoran Yu, Jingzhou Xu
Praxa is an evidence‑bound harness for governed AI agent execution that explicitly represents states such as proposal, authority, dispatch, verified external effect, and promotion through deterministic admission, brokered execution, external read‑back, reconciliation, and reviewed promotion. The authors report four evidence lanes: a repository‑local audit passing all unit and Workerd tests; a pilot on 12 curated tasks where both baseline and reliability‑layer arms passed 17 of 36 trials; a coordination‑proxy comparison where both baseline and a source‑authored candidate completed all 180 trials with equal accuracy but the candidate used fewer tokens and steps; and deployed source/configuration evidence showing bounded reflection, recall accounting, memory compilation, and tool‑health paths. None of the evidence demonstrates superiority in security, safety, or user benefit.
whyItMatters:"Praxa provides a testable architecture that makes authority‑to‑effect transitions explicit, offering a framework for verifying AI agent behavior, though current evidence does not prove improved security or performance."
By Stefan G. Creadore
The paper investigates the reliability of software produced by agentic AI by comparing AI-generated versions of ten well-known Linux utilities to their human-written counterparts. Using fuzz testing (both black-box and coverage-guided AFL++), the authors find that AI-generated code is often as reliable or more reliable than the latest human versions, with fewer memory errors but a higher incidence of hangs. The study emphasizes that robust AI-generated software requires careful prompting, skilled human oversight, and that the AI workflow can serve as a cost-effective specification for sustainable code.
By Ayesha Shafique, Barton P. MIller, Elisa R. Heymann