mlsanityData to 14 Oct 2026

Cognitive Kernel-Pro: A Framework for Deep Research Agents and Agent Foundation Models Training

arXiv 2508.00414 ↗First posted 1 Aug 2025cs.AI cs.CLcited by 42

On the map: LLM agents and tool use → Web, search, and research agents

Abstract (arXiv)

General AI Agents are increasingly recognized as foundational frameworks for the next generation of artificial intelligence, enabling complex reasoning, web interaction, coding, and autonomous research capabilities. However, current agent systems are either closed-source or heavily reliant on a variety of paid APIs and proprietary tools, limiting accessibility and reproducibility for the research community. In this work, we present Cognitive Kernel-Pro, a fully open-source and (to the maximum extent) free multi-module agent framework designed to democratize the development and evaluation of advanced AI agents. Within Cognitive Kernel-Pro, we systematically investigate the curation of high-quality training data for Agent Foundation Models, focusing on the construction of queries, trajectories, and verifiable answers across four key domains: web, file, code, and general reasoning. Furthermore, we explore novel strategies for agent test-time reflection and voting to enhance agent robustness and performance. We evaluate Cognitive Kernel-Pro on GAIA, achieving state-of-the-art results among open-source and free agents. Notably, our 8B-parameter open-source model surpasses previous leading systems such as WebDancer and WebSailor, establishing a new performance standard for accessible, high-capability AI agents. Code is available at https://github.com/Tencent/CognitiveKernel-Pro

Key facts. Cited by 42 papers in the map (data through 14 Oct 2026). Placed in Web, search, and research agents, one of 16 directions in LLM agents and tool use. Builds on 23 of its 36 references that are in the map. First posted on arXiv 1 Aug 2025.

Builds on

23 of 36 references.

Closest ancestors · 6

Same direction, newest first

  1. 2025WebCoT: Enhancing Web Agent Reasoning by Reconstructing Chain-of-Thought in Reflection, Branching, and Rollback26 May
  2. 2025WebEvolver: Enhancing Web Agent Self-Improvement with Coevolving World Model23 Apr
  3. 2025WebRollback: Enhancing Web Agents with Explicit Rollback Mechanisms16 Apr
  4. 2025Search-o1: Agentic Search-Enhanced Large Reasoning Models9 Jan
  5. 2024OpenWebVoyager: Building Multimodal Web Agents via Iterative Real-World Exploration, Feedback and Optimization25 Oct
  6. 2024WebVoyager: Building an End-to-End Web Agent with Large Multimodal Models25 Jan

Other references · 17

  1. 2023GAIA: a benchmark for General AI Assistantscited by 1,128
  2. 2025Group-in-Group Policy Optimization for LLM Agent Trainingcited by 497
  3. 2024Scaling Synthetic Data Creation with 1,000,000,000 Personascited by 417
  4. 2025R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learningcited by 336
  5. 2025WebThinker: Empowering Large Reasoning Models with Deep Research Capabilitycited by 335
  6. 2025WebWalker: Benchmarking LLMs in Web Traversalcited by 175
  7. 2023TACO: Topics in Algorithmic COde generation datasetcited by 160
  8. 2025OWL: Optimized Workforce Learning for General Multi-Agent Assistance in Real-World Task Automationcited by 131
  9. 2024TableBench: A Comprehensive and Complex Benchmark for Table Question Answeringcited by 128
  10. 2025WebShaper: Agentically Data Synthesizing via Information-Seeking Formalizationcited by 121
  11. 2025Scaling Test-time Compute for LLM Agentscited by 63
  12. 2021RiddleSense: Reasoning about Riddle Questions Featuring Linguistic Creativity and Commonsense Knowledgecited by 61
5 more
  1. 2025AutoAgent: A Fully-Automated and Zero-Code Framework for LLM Agentscited by 52
  2. 2024DOCBENCH: A Benchmark for Evaluating LLM-based Document Reading Systemscited by 37
  3. 2023LogiCoT: Logical Chain-of-Thought Instruction-Tuningcited by 31
  4. 2024TapeAgents: a Holistic Framework for Agent Development and Optimizationcited by 13
  5. 2024Cognitive Kernel: An Open-source Agent System towards Generalist Autopilotscited by 10

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 17; 2026: 25.

42 papers.

Sort:

By area:

  1. 2025MemEvolve: Meta-Evolution of Agent Memory Systemscited by 95
  2. 2025MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scalingcited by 81
  3. 2025Scaling Agents via Continual Pre-trainingcited by 40
  4. 2025Agentic Entropy-Balanced Policy Optimizationcited by 40
  5. 2026OpenResearcher: A Fully Open Pipeline for Long-Horizon Deep Research Trajectory Synthesiscited by 37
  6. 2026AOrchestra: Automating Sub-Agent Creation for Agentic Orchestrationcited by 35
  7. 2026Inference-Time Scaling of Verification: Self-Evolving Deep Research Agents via Test-Time Rubric-Guided Verificationcited by 22
  8. 2025DeepWideSearch: Benchmarking Depth and Width in Agentic Information Seekingcited by 21
  9. 2025Open Data Synthesis For Deep Researchcited by 16
  10. 2026EvoRoute: Experience-Driven Self-Routing LLM Agent Systemscited by 16
  11. 2026SmartSearch: Process Reward-Guided Query Refinement for Search Agentscited by 14
  12. 2026DynaWeb: Model-Based Reinforcement Learning of Web Agentscited by 14
30 more
  1. 2026Verified Critical Step Optimization for LLM Agentscited by 14
  2. 2026How to Allocate, How to Learn? Dynamic Rollout Allocation and Advantage Modulation for Policy Optimizationcited by 14
  3. 2026OffSeeker: Online Reinforcement Learning Is Not All You Need for Deep Research Agentscited by 12
  4. 2025WebAggregator: Enhancing Compositional Reasoning Capabilities of Deep Research Agent Foundation Modelscited by 10
  5. 2026Robust Tool Use via Fission-GRPO: Learning to Recover from Execution Errorscited by 10
  6. 2026SPPO: Sequence-Level PPO for Long-Horizon Reasoning Taskscited by 9
  7. 2026Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Applicationcited by 8
  8. 2025Automatically Generating Web Applications from Requirements Via Multi-Agent Test-Driven Developmentcited by 7
  9. 2025Understanding DeepResearch via Reportscited by 7
  10. 2025Towards Self-Evolving Benchmarks: Synthesizing Agent Trajectories via Test-Time Exploration under Validate-by-Reproduce Paradigmcited by 5
  11. 2025The Cognitive Bandwidth Bottleneck: Shifting Long-Horizon Agent from Planning with Actions to Planning with Schemascited by 5
  12. 2025DeepPlanner: Scaling Planning Capability for Deep Research Agents via Advantage Shapingcited by 5
  13. 2025HSCodeComp: A Realistic and Expert-level Benchmark for Deep Search Agents in Hierarchical Rule Applicationcited by 5
  14. 2026A-MapReduce: Executing Wide Search via Agentic MapReducecited by 5
  15. 2025Efficient Tree-Structured Deep Research with Adaptive Resource Allocationcited by 4
  16. 2026BAPO: Boundary-Aware Policy Optimization for Reliable Agentic Searchcited by 3
  17. 2026Training LLM Agents for Spontaneous, Reward-Free Self-Evolution via World Knowledge Explorationcited by 3
  18. 2026Code2Math: Can Your Code Agent Effectively Evolve Math Problems Through Exploration?cited by 2
  19. 2026DeepSearch-World: Self-Distillation for Deep Search Agents in a Verifiable Environmentcited by 2
  20. 2026Scaling the Scaling Logic: Agentic Meta-Synthesis of Logic Reasoningcited by 1
  21. 2026DRBENCHER: Can Your Agent Identify the Entity, Retrieve Its Properties and Do the Math?cited by 1
  22. 2026SciResearcher: Scaling Deep Research Agents for Frontier Scientific Reasoningcited by 1
  23. 2026Mitigating Context Interference for Reliable and Efficient Search Agentscited by 1
  24. 2025Anemoi: A Semi-Centralized Multi-agent System Based on Agent-to-Agent Communication MCP server from Coral Protocolcited by 0
  25. 2025ACADREASON: Exploring the Limits of Reasoning Models with Academic Research Problemscited by 0
  26. 2025Needle in the Web: A Benchmark for Retrieving Targeted Web Pages in the Wildcited by 0
  27. 2026Mind DeepResearch Technical Reportcited by 0
  28. 2026Placing Puzzle Pieces Where They Matter: A Question Augmentation Framework for Reinforcement Learningcited by 0
  29. 2026SciLens: Multi-modal Scientific Claim Verification with Agentic Entailment and Groundingcited by 0
  30. 2026OPERA: A Unified Omnimodal Progressive Spatio-Temporal Reasoning Agent for Referring Video Segmentationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026ElephantBroker: A Knowledge-Grounded Cognitive Runtime for Trustworthy AI Agents
  2. 2025ADRA-Bank: A Modular Benchmark for Academic Deep Research Agents
  3. 2026ResearchGym: Evaluating Language Model Agents on Real-World AI Research
  4. 2026MiroFlow: Towards High-Performance and Robust Open-Source Agent Framework for General Deep Research Tasks
  5. 2025Deep Research Agents: A Systematic Examination And Roadmap
  6. 2025Deep Research Bench: Evaluating AI Web Research Agents
  7. 2026Messier: A High-Resolution Corpus for Cross-Benchmark Agent Evaluation
  8. 2026K-Bench: measuring model performance on real scientific agent requests