MCPcopy Create free account
hub / github.com/Yeti-791/Tsec-Hackathon

github.com/Yeti-791/Tsec-Hackathon @main

Chat with this repo
repository ↗ · DeepWiki ↗ · + Follow
28 symbols 35 edges 1 files 28 documented · 100% updated 42d ago★ 768

Browse by type

Functions 28 Types & classes 0
What it actually does AI analysis from the code graph — generated when you open this
loading…
README

🤖Tsec-Hackathon - 腾讯云智能渗透黑客松

Tencent Cloud Intelligent Penetration Agent Hackathon

Clipboard_Screenshot_1778561225

本仓库为腾讯云智能渗透黑客松官方赛事资源仓,存放黑客松获奖团队答辩PPT、赛事物料及相关技术资料,同时整合赛事前二十优秀团队的开源项目仓库地址,打造智能渗透Agent领域一站式导航页面,助力网络安全从业者学习、交流与创新。

This repository is the official event resource warehouse for Tencent Cloud Intelligent Penetration Hackathon, which mainly stores the defense PPTs, event materials and related technical documents of the winning teams of the first Intelligent Penetration Hackathon. It also integrates the open source project repository addresses of the top 20 outstanding teams in the event to build a one-stop navigation page for the field of Intelligent Penetration Agent, helping cybersecurity practitioners learn, communicate and innovate.

📖赛事介绍 / Event Introduction

腾讯云智能渗透黑客松由腾讯云鼎实验室主办,是国内 首个聚焦 LLM 智能体全流程自动化渗透 的顶级专业赛事。赛事已连续成功举办两届,持续引领「AI + 安全」前沿技术探索与高端安全人才培养方向。赛事秉持 铸刃止戈、以智御危 理念,深度推动大模型与网络安全场景融合创新,探索智能渗透技术落地实践路径,同时面向产学研各界搭建高端 AI 安全竞技舞台,为行业持续输送顶尖 AI 安全实战人才。

Tencent Cloud Intelligent Penetration Hackathon, hosted by Tencent Cloud Yunding Lab, is China’s first top-tier professional competition focusing on full-process automated penetration based on LLM agents.Successfully held for two consecutive sessions, the event keeps spearheading cutting-edge exploration in AI + cybersecurity and the cultivation of high-end security talents.Upholding the philosophy of Forging Blades to Defend Threats, Guarding Risks with Intelligence, the competition deeply drives the integrated innovation of large models and cybersecurity scenarios, and explores the practical implementation path of intelligent penetration technologies.It also builds a high-end AI security arena for industry, academia and research communities, continuously delivering top practical AI security talents to the industry.

两届赛事累计汇聚800 + 支战队、千余名顶尖选手,产出 20 套顶尖智能渗透技术框架,形成 “赛事实践 - 技术沉淀 - 开源共享 - 行业赋能” 的良性循环。本板块收录两届赛事线上排名前十优秀团队的开源项目仓库导航,涵盖智能渗透 Agent 的核心设计思路、技术实现细节、实战攻防方案,完整呈现从初代可行性验证到高阶复杂场景落地的技术演进路径,是学习智能渗透技术、掌握 AI 攻防核心能力的权威参考资源。

Over the two competitions, more than 800 teams and over 1,000 top participants took part. Twenty state-of-the-art intelligent penetration frameworks were delivered, building a virtuous cycle featuring competition practice, technical accumulation, open-source sharing and industry empowerment. This section provides links to open-source repositories of the top 10 teams from the online stages of both events. It includes core design concepts, technical implementation details and practical attack and defense solutions for intelligent penetration agents, and fully presents the technical evolution from initial feasibility verification to deployment in sophisticated scenarios. It serves as an authoritative reference for learning intelligent penetration technologies and mastering core AI offensive and defensive capabilities.

🔥两届赛事高Star作品

⚔️第二届前20优秀团队项目 / Top 20 Teams Project Navigation

比赛时间:2026年4月(受技术发展速度影响,能力仅代表该时间节点)

本板块为赛事线上排名前二十的优秀团队开源项目仓库导航,涵盖智能渗透Agent的核心设计思路、技术实现与实战方案,是学习智能渗透技术的核心参考资源,排名按赛事最终成绩排序:

This section provides a navigation list of open-source project repositories from the top 20 teams in the online competition rankings. It covers the core design concepts, technical implementations and practical solutions of intelligent penetration Agents. Serving as a key reference resource for learning intelligent penetration technologies, the list is sorted by the final competition results.
排名 战队名 核心亮点 答辩PPT 视频 开源链接
1 ai小分队 Manager+Solver+Observer三层解耦。Observer旁路监督不干预执行,RTK Rewrite三层压缩解决上下文腐败,Ralph-Loop系统状态约束结束判定,7模型并行竞争上岗 《Adaptive Architecture for Pentest Agents》 播放 BreachWeave
3 Bytex 黑板系统+蚁群算法+涌现行为,平等Worker动态任务,反对预定义角色分工,认为其是人类局限的投影。全场唯一AK,7692元成本 《Cairn AI 从渗透测试到通用问题的求解》 播放 Cairn
7 For Future 使用纯自然语言FSM执行引擎;秉持Less Than Nothing哲学,零领域知识是主动排除而非遗漏,为LLM涌现留出结构化空位;Coordinator/P2P/Craft三种组织模式让AI自组织,而非硬编码角色分工 《Less Than Nothing》 播放 aide-for-pentest
17 爱吃大红袍茶叶蛋 - - - LingXi
18 青松 - - - llmnor
19 云南大学/西南石油大学 - - - cloudever_tecent_penetration
20 别用假装努力掩盖懒惰 - - - hackathon-pentest

第二届获奖项目团队特色简介 / Team Feature Introduction

获奖团队答辩PPT下载

Clipboard_Screenshot_1778492892

📈赛事复盘分析 / Event Review and Analysis

[Freebuf]智能攻防元年:渗透测试Agent迎来大考,AI如何从“能打”走向“可控”

[看雪学苑]前沿观察 赛事纪实:从腾讯云黑客松,洞见智能体时代的攻防新格局

重新定义智能渗透:TCH优秀开源项目CyberStrikeAI 技术解析与行业观察

两届腾讯云黑客松总结与分析

腾讯云智能渗透挑战赛 Agent 架构分析-SecureNexusLab

两届TCH之后 —— AI 渗透测试 Agent 的 Harness 工程演进、防御与我的思考

再论 AI 渗透测试 Agent 的减法哲学

AI自动化渗透是否到来?--TCH分析

国内最强 AI 渗透测试 Agent —— TCH·腾讯云黑客松第二届智能渗透挑战赛 唯一 AK 战队复盘

[TCH]腾讯云黑客松 第二届智能渗透挑战赛复盘

无径之径:Cairn AI 从渗透测试到通用问题的求解

Self-Evolving Kill Chain:Agent自适应进化与实战

Less Than Nothing - 从本质的本质出发,AI攻防的另一条路

记一次腾讯云智能渗透挑战赛复盘

浪潮将至--腾讯智能渗透赛冠军之夜回顾

AI 时代的旁观者 - 第二届腾讯云黑客松智能渗透挑战赛赛后记录

从零构建 AI 渗透测试 Agent:TCH 智能渗透黑客松实战复盘

TCH智能渗透赛: 你的下一个渗透 AI 为什么一定要是渗透 AI?

📊赛事模型网关日志报表 / LLM_Gateway_Public_Report

所有战队完整对话记录:https://challenge.zc.tencent.com/teams

战队模型使用情况分析报表: https://docs.qq.com/sheet/DUXBXY0R4VVNoaER1?nlc=1&tab=000001

Clipboard_Screenshot_1779957960

更新:之前名次有误

⚔️首届前20优秀团队项目 / Top 20 Teams Project Navigation

比赛时间:2025年11月(受技术发展速度影响,能力仅代表该时间节点)
排名 战队名 核心亮点 开源链接
2 xjtuHunter 基于场景感知的下一代黑盒渗透方案 xjtuHunter
3 BinX 基于状态感知与因果推理的自主渗透测试智能体 LuaN1aoAgent
4 Antix 100行代码,无需调优,完全由人工智能驱动 tinyctfer
6 NeuroSploit 具备认知能力的渗透智能体,AI自主规划与深度理解 Neuro-Sploit
7 ai小分队 AI 渗透的“蜂群思维” xbow-competition
8 D@wnEdg3 Cruiser: CTF Agent实现探索,实战攻防能力智能化演进 Cruiser Agent
9 yhy ChYing Agent:人机协作下的CTF自动化实践 CHYing-Agent
10 sickhack finds a way,AI驱动的安全自动化工具研发 SickHackShark
15 华科金银湖 基于CrewAI的ReAct架构,重方法论引导的多Agent newmapta
16 瀚海星云 多Agent安全测试系统 sub-agent-autopt
17 小白战队 基于go语言构建,集成了100多种安全工 CyberStrikeAI
18 HRP(Nepnep) 全异步架构、三层智能决策、50+攻击知识库 H-Pentest
25 基米牌南北绿豆好吃吗 专为CTF设计的可扩展AI Agent BUUCTF_Agent
35 中传C1JC战队 基于OODA的迭代式笔记本问题求解自主代理 AgentNote
> #### 获奖团队答辩PPT下载
>
#### 首届特色团队简介 / Team Feature Introduction
  1. xjtuHunter(西安交通大学):由网络空间安全学院师生组成,深耕智能攻击检测、自动化漏洞挖掘,研究成果发表于ASE、NDSS等国际顶级会议。

  2. Antix(ChainReactors):致力于构建AI原生进攻性安全基座,通过先进AI Agent工程打造下一代攻防作战指挥平台。

  3. BinX(广州大学):秉承方滨兴院士育人理念,承担多项国家级重大课题及大型赛事网络安全保障,深耕智能攻防与自动化渗透。

赛事复盘分析 / Event Review and Analysis

四川大学;清华大学等:黑客还是幻觉?基于大语言模型的自动化渗透测试全面分析

渗透成功率超94%!长亭科技AI自主渗透智能体再获“腾讯云黑客松”决赛冠军

Intent Is All You Need (for agent)

7天Top 9:我如何让 Claude 手搓一个全自动 CTF 选手

The above teams represent the top level of domestic intelligent penetration Agent technology, covering core directions such as multi-agent collaboration, autonomous planning of LLM agents, and scenario-aware black-box penetration, and have important reference value for the research and development of intelligent penetration technology.

🏆赛事奖励 / Event Rewards

本次黑客松设置丰厚的现金奖励与开源贡献奖,鼓励AI安全技术创新与开源共享:

  • 一等奖(第1名):¥60000

  • 二等奖(第2~3名):¥40000

  • 三等奖(第4~6名):¥20000

  • 优秀奖(第7~10名):¥5000

  • 开源贡献奖:前20名开源贡献者获腾讯颁发的荣誉证书及奖杯

The event takes "AI-driven penetration" as the core orientation, and strictly restricts manual penetration operations, creating a fair, transparent and technology-oriented competition atmosphere.

🤝合作伙伴 / Collaborator

Clipboard_Screenshot_1779868365

📚学习资料 / Learning assets

Offensive AI Agentic 全景:开源项目 / 论文 / Benchmark / 商业产品 一览
本文档系统整理了 AI 渗透测试 / LLM 红队 / 自主攻击 Agent / 漏洞挖掘 领域的开源项目、学术论文、能力评测 Benchmark 与国内外商业化解决方案,旨在帮助研究者、安全工程师与企业安全决策者快速建立领域全景认知。

This document curates open-source projects, academic papers, capability benchmarks, and commercial solutions (international & China) in the field of AI penetration testing, LLM red teaming, autonomous offensive agents, and vulnerability discovery, aimed at helping researchers, security engineers, and enterprise decision-makers quickly form a holistic view.

https://github.com/Yeti-791/Awesome-Offensive-AI-Agentic-Landscape

💝致谢 / Acknowledgements

由衷感谢腾讯安全云鼎实验室、广州大学、新安盟为本届智能渗透黑客松在策划、运营及技术保障方面提供的大力支持;感谢全体参赛团队积极开展技术创新并乐于开源分享,助力 AI 大模型与智能渗透技术深度融合;同时感谢北京大学、清华大学、华中科技大学、西安交通大学、国防科技大学、南洋理工大学、成都信息工程大学、中国电科院、天翼安全、京东科技、中国电信、长亭科技、绿盟科技、安恒信息等各大高校、企业与科研机构对赛事的参与和鼎力支持;最后向所有为网络安全 AI 技术生态建设付出努力的开发者与社区致以诚挚谢意。

_We sincerely appreciate Tencent Security Yunding Lab, Guangzhou University and Xin'an Alliance for their core support in the planning, operation and technical guarantee of the Intelligent Penetration Hackathon. We thank all participating teams for their technological innovation and open-source sharing, which promote the integrated development of large language models and intelligent penetration technologies. We also extend our gratitude to

Core symbols most depended-on inside this repo

browse all functions →

Shape

Function 28

Languages

Python100%

Modules by API surface

第二届智能渗透黑客松/零界论坛工具包/mcp_server.py28 symbols

For agents

$ claude mcp add Tsec-Hackathon \
  -- python -m otcore.mcp_server <graph>

⬇ download graph artifact

Ask about this repo answers extend the page