{"full_name":"vxcontrol/pentagi","owner":"vxcontrol","name":"pentagi","url":"https://github.com/vxcontrol/pentagi","description":"Fully autonomous AI Agents system capable of performing complex penetration testing tasks","homepage":"https://pentagi.com","topics":[],"language":"Go","license":"MIT","avatar":"https://avatars.githubusercontent.com/u/39194251?v=4","created_at":"2025-01-06T22:22:48Z","pushed_at":"2026-10-22T23:26:13Z","metrics":{"stars":33839,"forks":5313,"watchers":180,"issues":163,"prs":306,"releases":10,"issues_open":null,"issues_closed":null,"prs_open":null,"prs_closed":null,"prs_merged":null,"discussions":5},"growth":{"day":{"stars":90,"forks":26,"base_day":"2026-11-09"},"week":{"stars":786,"forks":148,"base_day":"2026-11-03"},"month":{"stars":8577,"forks":2068,"base_day":"2026-10-11"}},"history":{"dates":["2026-09-06","2026-09-07","2026-09-08","2026-09-09","2026-09-10","2026-09-11","2026-09-12","2026-09-13","2026-09-14","2026-09-15","2026-09-16","2026-09-17","2026-09-18","2026-09-19","2026-09-20","2026-09-21","2026-09-22","2026-09-23","2026-09-24","2026-09-25","2026-09-26","2026-09-27","2026-09-28","2026-09-29","2026-09-30","2026-10-01","2026-10-02","2026-10-03","2026-10-04","2026-10-05","2026-10-06","2026-10-07","2026-10-08","2026-10-09","2026-10-10","2026-10-11","2026-10-12","2026-10-13","2026-10-14","2026-10-15","2026-10-16","2026-10-17","2026-10-18","2026-10-19","2026-10-20","2026-10-21","2026-10-22","2026-10-23","2026-10-24","2026-10-25","2026-10-26","2026-10-27","2026-10-28","2026-10-29","2026-10-30","2026-10-31","2026-11-01","2026-11-02","2026-11-03","2026-11-04","2026-11-09","2026-11-10"],"stars":[21994,22017,22032,22071,22104,22122,22165,22223,22251,22270,22281,22324,22509,22564,22621,22684,22971,23314,24396,24486,24577,24660,24707,24748,24834,24867,24905,24928,24959,24991,25071,25110,25172,25199,25219,25262,25278,25307,25342,25394,25432,26522,28089,29154,30407,30709,30882,31306,31516,31731,32016,32161,32288,32497,32576,32652,32736,32804,33053,33167,33749,33839],"forks":[2920,2919,2923,2931,2941,2944,2951,2955,2955,2955,2955,2956,2975,2978,2989,2998,3028,3052,3127,3135,3149,3169,3177,3179,3196,3202,3207,3208,3208,3210,3222,3232,3240,3239,3242,3245,3244,3244,3254,3267,3276,3688,4163,4380,4554,4671,4744,4951,4958,4885,4952,4989,5004,5054,5073,5088,5102,5124,5165,5180,5287,5313],"watchers":[130,131,131,133,136,136,136,137,137,138,138,138,140,140,141,140,141,143,145,146,145,147,148,148,148,148,148,148,147,148,148,148,149,149,150,150,150,150,150,150,150,152,158,160,160,163,165,171,169,170,174,174,171,171,172,171,172,172,174,175,180,180],"issues_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null],"prs_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null]},"boards":{"weekly":147,"monthly":70,"weekly_rate":156,"monthly_rate":71},"same_owner":[],"issues":[{"number":432,"title":"[Bug]: Cannot login on fresh install","comments":8,"created_at":"2026-09-30T16:27:14Z"},{"number":449,"title":"[Bug]: AI Agents Settings cannot be saved","comments":5,"created_at":"2026-10-15T09:54:30Z"},{"number":455,"title":"[Enhancement]:Allow MVP and potential Skill features to access external custom knowledge bases","comments":4,"created_at":"2026-10-17T01:48:17Z"},{"number":403,"title":"[Bug]: Resources page freezes with large resource libraries","comments":4,"created_at":"2026-09-06T12:46:00Z"},{"number":71,"title":"[Enhancement]: scope of work menu","comments":4,"created_at":"2025-10-25T13:23:40Z"},{"number":391,"title":"[Bug]: In some cases, retrieving flow details may hang","comments":3,"created_at":"2026-08-12T03:50:17Z"},{"number":361,"title":"[Bug]: Triple backticks inside code blocks break markdown rendering","comments":3,"created_at":"2026-06-29T22:30:28Z"},{"number":486,"title":"[Security] a failed sandbox probe still gives the agent the Docker daemon, without verifying that it blocks host escapes","comments":2,"created_at":"2026-11-04T09:17:15Z"}],"discussions":[{"number":315,"title":"Custom LLM Provider (Groq) + Ollama CPU timeout — configuration questions","comments":2,"category":"Q&A","created_at":"2026-05-21T05:21:47Z"},{"number":229,"title":"Git History Rewrite - Licensing Resolution and Path Forward","comments":0,"category":"Announcements","created_at":"2026-03-29T17:54:19Z"},{"number":217,"title":"Feedback from internal security team assessment — Is this a fair evaluation of pentagi's current capabilities?","comments":1,"category":"General","created_at":"2026-03-19T15:13:26Z"},{"number":130,"title":"EXPORT METHOD","comments":1,"category":"Q&A","created_at":"2026-02-23T01:33:23Z"},{"number":115,"title":"great work","comments":3,"category":"General","created_at":"2026-02-21T17:19:09Z"}],"latest_release":{"name":"2.2.0","tag":"v2.2.0","published_at":"2026-10-05T06:59:50Z","body":"---\r\n\r\n<div align=\"center\">\r\n\r\n**PentAGI 2.2 — Any Modern LLM, a Contained Sandbox, Unified Web Search, and Multi-Instance Deployment**\r\n\r\nThis release supports the current generation of LLMs with a tested configuration for every provider, closes CVE-2026-14784 with rebuilt sandbox isolation, replaces seven search tools with one intent-driven `web_search`, and lets several installations share one set of backing services.\r\n\r\n[![Discord](https://img.shields.io/badge/Discord-7289DA?logo=discord&logoColor=white)](https://discord.gg/2xrMh7qX6m)⠀[![Telegram](https://img.shields.io/badge/Telegram-2CA5E0?logo=telegram&logoColor=white)](https://t.me/+Ka9i6CNwe71hMWQy)\r\n\r\nTo get an Enterprise Edition license, sign up at [console.pentagi.com](https://console.pentagi.com) — new users get one free license there.\r\n\r\n</div>\r\n\r\n---\r\n\r\n### Major Features\r\n\r\n#### **Any Modern LLM, With a Tested Configuration**\r\n\r\nLangChainGo moves from `v0.1.14-update.5` to `update.8`, and PentAGI supports the current model generations without workarounds.\r\n\r\n- **Reasoning the way each vendor expects it** — adaptive thinking, effort levels, thinking budgets, and a real \"off\" are sent in each model family's native form. Combinations a vendor would reject are caught or redirected before the request instead of coming back as a 400, signed reasoning replays intact across agent turns, and Settings offers only the controls the selected model honors.\r\n- **Three new providers** — **MiniMax**, **Mistral**, and **xAI**, for 13 provider types in total. Any other OpenAI-compatible endpoint works through the custom provider, which now also calls **Azure OpenAI** deployments (`LLM_SERVER_API_TYPE=azure`).\r\n- **Tested, not just listed** — every built-in provider, plus example configurations for OpenRouter, DeepInfra, Ollama Cloud, self-hosted vLLM, and others, ships with a `ctester` report in `examples/tests/` giving the success rate per agent role.\r\n- **Current catalogues** — GPT-6, Claude Opus 5.x and Fable 5.x, Gemini 3.x, Qwen 3.8, GLM 5.3, DeepSeek V4, and Grok 4.7, with prices checked against vendor pages. Agent chains are compacted to the model's known context window before each call.\r\n- **Provider failover** — `LLM_FALLBACK_PROVIDER` repeats a failed model call on a second configured provider and records the switch in the trace.\r\n- **Anthropic without a long-lived key** — Workload Identity Federation exchanges a Kubernetes, GitHub Actions, or other OIDC token for a short-lived Claude token.\r\n\r\n#### **Sandbox Security and CVE-2026-14784**\r\n\r\n[CVE-2026-14784](https://nvd.nist.gov/vuln/detail/CVE-2026-14784) affects PentAGI up to 2.1.0: with `DOCKER_INSIDE=true`, as the shipped `.env.example` set it, the host Docker socket was mounted into every sandbox, so an agent steered by prompt injection could take over the host. Prompt injection cannot be closed completely — resistance depends largely on the model's instruction following and generalization — so this release concentrates on containing whatever runs in the sandbox.\r\n\r\n- **Least privilege** — worker containers drop every capability and add back an explicit allow-list: `MKNOD` is withheld (the main known escape path), `SYS_PTRACE` is added for debuggers, and a process limit guards against fork bombs.\r\n- **No host socket** — `DOCKER_INSIDE` defaults to `false`, and nested Docker now requires `DOCKER_INSIDE_HOST`, a daemon separate from the one running PentAGI. The host socket is no longer mounted automatically.\r\n- **Startup attestation** — with `DOCKER_INSIDE_POLICY_TESTS=true`, PentAGI launches a worker exactly as a flow would and verifies from inside it that the daemon is not its own, 27 host-escape requests are refused, and 7 legitimate operations work. If a check fails, agents get no Docker access and PentAGI keeps running.\r\n- **Hardened worker node** — the [worker node guide](https://github.com/vxcontrol/pentagi/blob/main/examples/guides/worker_node.md) ships a Docker-in-Docker kit: a fail-closed OPA policy, a sec"},"readme":true,"readme_zh":false,"news":[{"title":"Pentagi: Autonomous AI Agents for complex penetration testing tasks","url":"https://github.com/vxcontrol/pentagi","hn":"https://news.ycombinator.com/item?id=47114192","source":"Hacker News","lang":"en","date":"2026-02-22","points":2,"comments":0},{"title":"PentAGI: Open-source autonomous AI penetration testing system","url":"https://www.helpnetsecurity.com/2026/04/22/pentagi-autonomous-ai-penetration-testing/","source":"Help Net Security","lang":"en","date":"2026-04-21"},{"title":"PentAGI: The Open-Source AI That Could Transform Cybersecurity Forever (and Its Hidden Risks)","url":"https://www.tech2geek.net/pentagi-the-open-source-ai-that-could-transform-cybersecurity-forever-and-its-hidden-risks/","source":"tech2geek","lang":"en","date":"2026-03-22"},{"title":"Autonomous AI red-team tool PentAGI released as open source","url":"https://www.digitaltoday.co.kr/en/view/42188/pentagi-autonomous-ai-red-team-tool-open-source-release","source":"디지털투데이","lang":"en","date":"2026-03-21"},{"title":"AI Hacks Autonomously! \"PentAGI\" is Changing the Security Industry! Autonomous AI Penetration Testing Integrating Over 20 Tools (February 2026) | AI Study Boost","url":"https://note.com/calm_ruff735/n/n101d96941f22?hl=en","source":"note","lang":"en","date":"2026-02-22"},{"title":"PentAGI – Automated AI-Powered Penetration Testing Tool that Integrates 20+ Security Tools","url":"https://cybersecuritynews.com/pentagi-penetration-testing-tool/","source":"cybersecuritynews","lang":"en","date":"2026-02-19"},{"title":"Beyond ARTEX: 45 Agentic Attack Frameworks, What They Hunt For, and How to Find Them","url":"https://news.google.com/rss/articles/CBMib0FVX3lxTE16VU80QUx1b1NyaG9CTkpJaXBLcWZhSkUyTlpKNUJoTnJsemJObFE1d29pVVpjeFlVejV5V3E2UXJEVzlfa2ZxUDUtR0g0aE1MOFZBRmlNb0hqU044YjZ2UExNcDl1YTB4MHNLSE9SZw?oc=5","source":"GitGuardian Blog","lang":"en","date":"2026-11-09"},{"title":"ARTEX and Friends: Enter Agentic Offensive Frameworks","url":"https://news.google.com/rss/articles/CBMirgFBVV95cUxOTHJrY05HcXR3TFd0YVF4MTVHQ3lvRDk0YnFDbzFCWDBqV0UyTWwxQXJVaUtSYTRIcFpPRDRLeE1lRjVrdVFRdndlQlpZWElTTmxOTE5FeDRoVHJIWFhPaFZvcTc4ODFHMEYxN0EyeFVpRG9MLW9yVnN0bmdkcl9PWkRYZHdqY3hMdHFaNjlkWnpqRFpDaFh5NGZPUzZ2a3FaSllrOHRpV25ORFItVlE?oc=5","source":"Google Cloud","lang":"en","date":"2026-10-30"},{"title":"当 AI 学会自主规划攻击路径，渗透测试还能靠人吗|pentagi|vxcontrol|渗透测试|多智能体_新浪新闻","url":"https://news.google.com/rss/articles/CBMiY0FVX3lxTE5xZm9QbTBIZkpPZmVTc0xDTW9qZV92Z1ZPV1B5aUJOdEFvZm1feHh0S19rVU5pNmxIYzl4aUpTWThSR0VIT1BJNXlIOGdSdkJxaWdDUUNPWGNqcGJHamdxQXMzUQ?oc=5","source":"新浪网","lang":"zh","date":"2026-10-13"},{"title":"9日，在开发者社区GitHub上查到的人工智能（AI）黑客工具很难简单归为某一种类型。它们细分为多种形态，从由AI自行探寻入侵路径的自主智能体，到统一管理多种黑客工具的平台，再到辅助攻击者作出判断的A..","url":"https://news.google.com/rss/articles/CBMiT0FVX3lxTE1zTmozcnBhdHkxN3lWUEwwckZfdVBmekVaMTBGNEFVN3VWNExOUFoydDJmVlllZGI5N0tpX2tXaloyU1QwMkRjNGlVQlQxQ0U?oc=5","source":"매일경제","lang":"zh","date":"2026-10-09"},{"title":"Security Patterns for Autonomous Agents: Lessons from Pentagi","url":"https://news.google.com/rss/articles/CBMikAFBVV95cUxOc1I1VDJfeXlmalFxSXF1NXFwbDZycnR2R3BWV3FjRElhTkZrMm85QWZJR2otNFhISW50Skpfa0RWenIxYWpydzRnUjEyS0swYWNxRTd2WjFQb3ppazZCLTRIVllXdUk3Yl91RXJkNEJQVlN4ZWsyR3N4d01DUUdEbjZjVEo3a052NVBPWENBUFI?oc=5","source":"SitePoint","lang":"en","date":"2026-02-23"}],"stale":false,"snapshot_at":"2026-11-10T09:35:25Z","history_start":"2026-09-06"}