主题 · Topics
主题 · Topics — 每日 Harness
按 Agent / Harness 主题跨刊期浏览全部内容——进入任一主题查看其全部条目并按日期筛选，或回到全部刊期按时间浏览。
共 7 个主题分组 · 21 个子主题 · 1146 条内容
1.1 推理与规划Reasoning & Planning 12 项
- PhoenixRepair: Rethinking Repair Strategy Exploration in Software Agents 2026-07-24arxiv.org
- Do Coding Agents Need Executable World Models, Simplification, and Verification to Solve ARC-AGI-3? 2026-07-21arxiv.org
- Can We Understand How Large Language Models Reason? 2026-07-13cacm.acm.org
- FirstResearch: Auditable Question Formation for LLM Scientific Discovery Agents 2026-07-09arxiv.org
- Understanding is the new bottleneck 2026-07-04geoffreylitt.com
1.2 测试时计算Test-time Compute 11 项
- Effort Router: Intelligent /effort selection per Claude turn 2026-07-21github.com
- Controlling Reasoning Effort in LLMs 2026-07-21magazine.sebastianraschka.com
- Doomed from the Start: Early Abort of LLM Agent Episodes via a Recall-Controlled Probe Cascade 2026-07-09arxiv.org
- SPORK: Self-Speculative Forking to Accelerate Agentic LLM Inference 2026-07-08arxiv.org
- Scaling with Confidence: Calibrating Confidence of LLMs for Adaptive Test Time Scaling 2026-07-04arxiv.org
2.1 Agent RL / 可验证奖励Agent RL / Verifiable Rewards 34 项
- ToolVerse: Unlocking Massive Environments and Long-Horizon Tasks for Agentic Reinforcement Learning 2026-07-21arxiv.org
- DSWorld: A Data Science World Model for Efficient Autonomous Agents 2026-07-21arxiv.org
- Agentic-DPO: From Imitation to Agentic Policy Optimization on Expert Trajectories 2026-07-15arxiv.org
- SWE-1.7 Reach Near GPT 5.5 and Opus Intelligence 2026-07-09cognition.com
- You Only Need 1 Layer for RLVR? 2026-07-08mail.bycloud.ai
2.2 蒸馏与压缩Distillation & Compression 9 项
- A Systematic Evaluation of Trajectory Data Curation for LoRA Fine-Tuning of Code Agents 2026-07-23arxiv.org
- cactus-compute/needle 2026-07-16github.com
- The Illusion of Equivalency 2026-07-11arxiv.org
- OPID: On-Policy Skill Distillation for Agentic Reinforcement Learning 2026-06-27arxiv.org
- Skill-Guided Continuation Distillation for GUI Agents 2026-06-19arxiv.org
2.3 自进化Self-Evolution 14 项
- Ingot 2026-07-23github.com
- Autoretrieval 2026-07-22github.com
- Agentic Synthesis against Counterexample-Supplemented Sketches 2026-07-21arxiv.org
- GPT-Red: Unlocking Self-Improvement for Robustness 2026-07-16openai.com
- Harness Engineering for Self-Improvement 2026-07-08lilianweng.github.io
2.4 合成数据与训练环境Synthetic Data & Environments 15 项
- Environment-free Synthetic Data Generation for API-Calling Agents 2026-07-25arxiv.org
- Masked Diffusion Language Models are Strong and Steerable Text-Based World Models for Agentic RL 2026-07-23arxiv.org
- PrimeIntellect-ai/verifiers 2026-07-16github.com
- Function-Aware Fill-in-the-Middle as Mid-Training for Coding Agent Foundation Models 2026-07-16arxiv.org
- Beyond Static Evaluation: Building Simulation Environments for Scalable Agentic Reinforcement Learning 2026-07-09arxiv.org
3.1 Agent 记忆Agent Memory 78 项
- Profile-Graph Memory for LLM Agents: Implicit Cross-Entity Traversal through Narrative Profiles 2026-07-25arxiv.org
- Marrow 2026-07-25github.com
- Fidelity Before Structure: Verbatim Chunks Beat Lossy Artifact Extraction in Long-Conversation LLM Memory 2026-07-25arxiv.org
- Supra Cognitive Modes: A Routed Architecture for Agent Memory 2026-07-24arxiv.org
- akitaonrails/ai-memory 2026-07-23github.com
3.2 上下文工程Context Engineering 80 项
- Why Software Factories Fail 2026-07-25github.com
- State Compression in Two-Agent LLM Relays 2026-07-24arxiv.org
- Prompt Caching in Agents 2026-07-24earendil.com
- yvgude/lean-ctx 2026-07-23github.com
- rtk-ai/rtk 2026-07-23github.com
3.3 检索与知识接地Retrieval / RAG 59 项
- Amdb 2026-07-25github.com
- Grounded Forge 2026-07-24github.com
- HKUDS/LightRAG 2026-07-23github.com
- tirth8205/code-review-graph 2026-07-21github.com
- Local-first CLI to make Obsidian vaults searchable for AI agents 2026-07-21github.com
4.1 工具使用Tool Use 79 项
- oraios/serena 2026-07-25github.com
- CockroachCrawler 2026-07-25github.com
- Browser Bridge 2026-07-25github.com
- Decode-Time Grammars 2026-07-24arxiv.org
- Browser Tools SDK 2026-07-23libretto.sh
4.2 技能系统Skills 56 项
- ibelick/ui-skills 2026-07-20github.com
- Nutlope/hallmark 2026-07-16github.com
- Dynamic Agent Skills: A Lifecycle Survey and Taxonomy of Evolving Skill Libraries 2026-07-15arxiv.org
- google-labs-code/stitch-skills 2026-07-13github.com
- dotnet/skills 2026-07-09github.com
4.3 协议与互操作Protocols & Interop 47 项
- Palmier Pro 2026-07-24github.com
- Claude Code Proxy 2026-07-24github.com
- PrefectHQ/fastmcp 2026-07-22github.com
- memorywire: A Vendor-Neutral Wire Format for Agent Memory Operations 2026-07-21arxiv.org
- Scalable LLM Agent Tool Access in the Cloud 2026-07-21arxiv.org
5.1 多智能体Multi-Agent 42 项
- block/buzz 2026-07-25github.com
- THU-MAIC/OpenMAIC 2026-07-25github.com
- Catalyst 2026-07-25github.com
- Syndicate 2026-07-24usesyndicate.org
- BDFL 2026-07-24usebdfl.com
5.2 工作流与控制流Workflows & Control 88 项
- alibaba/open-code-review 2026-07-25github.com
- Sigil 2026-07-25github.com
- Why Software Factories Fail 2026-07-24github.com
- CodeRescue: Budget-Calibrated Recovery Routing for Coding Agents 2026-07-24arxiv.org
- BatchDAG: LLM-Planned Execution Graphs for Scalable Ad-Hoc Analysis Over Enterprise Data 2026-07-24arxiv.org
6.1 框架与脚手架Frameworks & Scaffolds 147 项
- agegr/pi-web 2026-07-25github.com
- Rendi 2026-07-24github.com
- LLM Budget Cap 2026-07-24github.com
- Fleet 2026-07-24github.com
- MoonshotAI/kimi-code 2026-07-23github.com
6.2 执行环境与沙箱Execution & Sandboxing 59 项
- citrolabs/ego-lite 2026-07-25github.com
- AgentCgroup: Understanding and Controlling OS Resources of AI Agents 2026-07-25arxiv.org
- Pullrun 2026-07-24github.com
- Superserve 2026-07-23superserve.ai
- Serve-avd 2026-07-22github.com
6.3 可观测性与调试Observability & Debugging 53 项
- AgentTrails: Towards Trust and Reuse for Agentic Tasks 2026-07-24arxiv.org
- AgentPulse 2026-07-24prove-ai.github.io
- AgentDebugX 2026-07-24arxiv.org
- Deterministic Replay for AI Agent Systems 2026-07-23arxiv.org
- microsoft/AI-Engineering-Coach 2026-07-22github.com
7.1 基准Benchmarks 60 项
- DocOps: A Verifiable Benchmark for Autonomous Agents in Complex Document Operations 2026-07-25arxiv.org
- Alipay-PIBench: A Realistic Payment Integration Benchmark for Coding Agents 2026-07-25arxiv.org
- RECON: Benchmarking Agent Memory for Compositional Reasoning over Long Contexts 2026-07-23arxiv.org
- Memory Bench 2026-07-23github.com
- Lomekwi: Resource-Bounded Tool Discovery in LLM Agents 2026-07-23arxiv.org
7.2 评测方法Eval Methodology 77 项
- The First Known Runaway AI Agent - or a Very Bad Marketing Stunt? 2026-07-25simonwillison.net
- Silent Failures in Multimodal Agentic Search: A Diagnostic Taxonomy and Cross-Judge Evaluation 2026-07-25arxiv.org
- Guardrails as Scapegoats: Auditing Unfaithful Safety Refusals in Tool-Augmented LLM Agents 2026-07-25arxiv.org
- When JSON Is Not Enough 2026-07-24arxiv.org
- SAAG: Structured Agent Assessment and Grounding 2026-07-24arxiv.org
7.3 安全与攻防Security 113 项
- Stateful Guardrails for Multi-Turn LLM Systems: A Conversational Risk Accumulation Framework 2026-07-25arxiv.org
- NEXUS: Structured Runtime Safety for Tool-Using LLM Agents 2026-07-25arxiv.org
- JANUS: Foreseeing Latent Risk for Long-Horizon Agent Safety 2026-07-25arxiv.org
- ChannelGuard: Safe Models Do Not Compose into Safe Multi-Agent Systems 2026-07-25arxiv.org
- ChainWatch: A Kill Chain-Aligned Sequential Detection Framework for Multi-Step Attacks in MCP-Based AI Agent Systems 2026-07-25arxiv.org
7.4 对齐与治理Alignment & Governance 13 项
- Safety and alignment in an era of long-horizon models 2026-07-21openai.com
- ANCHOR: Automated Alignment Auditing for CLI Agents on Real-World Harm 2026-07-15arxiv.org
- Understand to participate 2026-07-03simonwillison.net
- AgentBound: Verifiable Behavioral Governance for Autonomous AI Agents 2026-07-02arxiv.org
- AgentWatch 2026-06-29agent-watch.dev