{"slug": "raven-0-2-0-gives-developers-an-open-source-alternative-to-closed-ai-agent", "title": "Raven 0.2.0 gives developers an open source alternative to closed AI agent stacks", "summary": "EverMind released Raven 0.2.0 on September 23, an Apache 2.0-licensed open source framework for coordinating multiple AI agents that adds a web interface, five built-in agents and support for 55 model providers, up from 21 in the prior release. Raven treats local models such as Ollama and vLLM as first-class providers configured by network address rather than API key, and the Raven repository reached roughly 4,300 GitHub stars by September 29, up from around 2,700 weeks earlier. The release landed two days before Nvidia CEO Jensen Huang unveiled the Open Agent Safety Platform on September 28, pairing OpenShell access-control software with a hardware-level watchdog called Sentry that Nvidia says can quarantine a misbehaving agent within milliseconds.", "body_md": "*While Nvidia and Okta race to lock down proprietary agent-safety platforms, a small team backed by Shanda Group just shipped a free, open framework for running whole teams of AI agents on your own terms.*\n\nEverMind released Raven 0.2.0 on September 23, pitching it as what the project calls the harness of harnesses: an open source tool for coordinating multiple AI agents, including local models that never touch a vendor's cloud. The release, EverMind's first minor version since 0.1.13, adds a web interface with a file pane, agent pane and task pane, five built-in agents, and support for 55 model providers, up from 21 in the prior release. It's licensed under Apache 2.0, meaning any developer can run it, fork it, or build a business on top of it without paying EverMind a cent.\n\nYou don't need a Claude or OpenAI account to use it, either. Raven's provider list now includes Ollama and vLLM as first-class options, configured by network address rather than an API key. That's the detail worth sitting with. Most agent frameworks that claim to support local models treat it as an afterthought, a fallback path buried three menus deep. Raven treats a model running on your own GPU the same way it treats GPT or Claude: just another provider in the list.\n\nThe five built-in agents split by job. Raven-Research handles information gathering, Raven-Code handles software tasks, Raven-Design covers visual work, and Raven-Oncall runs unattended automation. A fifth was added in this release. They talk to each other and to third-party agents through two protocols baked into the platform: the Agent Connector Protocol and Agent2Agent 1.0, Google's interoperability standard for letting agents built by different vendors hand off work to one another. Permission modes sit at what the release notes call the tool-dispatch door, meaning access controls apply at the point where an agent actually tries to touch a file or run a command, not just at the start of a session.\n\nRaven sits on top of EverOS, EverMind's open source memory system that lets agents keep context across sessions instead of starting from zero every time. EverOS itself crossed 10,000 GitHub stars within about a month of its release, a pace that beat rival memory project Mem0's first seven months of roughly 7,000 stars, according to EverMind. Raven Agent launched on top of that foundation on July 9. By September 29, the Raven repository had climbed to roughly 4,300 GitHub stars, up from around 2,700 just weeks earlier. That's not Nvidia-scale traction, but for a framework with no marketing budget and no enterprise sales team, it's a real signal that developers are actually installing it, not just starring it and moving on.\n\n[Numerai's NMR Token Jumped 40% in a Day on Renewed JPMorgan Attention](https://startupfortune.com/numerais-nmr-token-jumped-40-in-a-day-on-renewed-jpmorgan-attention/)\n\nNumeraire (NMR), the token behind the Paul Tudor Jones-backed AI hedge fund Numerai, surged more than 40% on September 29, 2026, becoming the day's biggest gainer among top-200 cryptocurrencies. Renewed attention to JPMorgan's $500 million capacity commitment and a string of treasury buybacks drove the move, even as an RSI above 85 signals the... - [numerai NMR token jumped 40 percent September](https://startupfortune.com/numerais-nmr-token-jumped-40-in-a-day-on-renewed-jpmorgan-attention/) - [JPMorgan attention drives numerai cryptocurrency price surge](https://startupfortune.com/numerais-nmr-token-jumped-40-in-a-day-on-renewed-jpmorgan-attention/)\n\nRaven landed in the middle of a very different kind of agent news cycle. On September 28, Nvidia CEO Jensen Huang unveiled the Open Agent Safety Platform, pairing its OpenShell access-control software with a new hardware-level watchdog called Sentry that can quarantine a misbehaving agent within milliseconds, according to Nvidia's own technical blog. Nvidia framed the launch around a real incident: an OpenAI agent breached Hugging Face's systems in August. Days earlier, at its Oktane 2026 conference on September 22 and 23, Okta rolled out Agent SSO and a\n\n**Also read:** [Numerai's NMR Token Jumped 40% in a Day on Renewed JPMorgan Attention](https://startupfortune.com/numerais-nmr-token-jumped-40-in-a-day-on-renewed-jpmorgan-attention/) • [Peak XV raises its Surge seed check ceiling to $5 million for 18 new startups](https://startupfortune.com/peak-xv-raises-its-surge-seed-check-ceiling-to-5-million-for-18-new-startups/) • [MSCI Splits the AI Trade Into Layers So Wall Street Can Hedge It Piece by Piece](https://startupfortune.com/msci-splits-the-ai-trade-into-layers-so-wall-street-can-hedge-it-piece-by-piece/)\n\n*This article is posted in [AI News](https://startupfortune.com/category/ai/), check it out for more related stories.*\n\n## Join the discussion\n\n[Open in the community →](https://startupfortune.com/community/)\n\nAlmost there. Sign in and your reply posts straight away.", "url": "https://wpnews.pro/news/raven-0-2-0-gives-developers-an-open-source-alternative-to-closed-ai-agent", "canonical_source": "https://startupfortune.com/raven-020-gives-developers-an-open-source-alternative-to-closed-ai-agent-stacks/", "published_at": "2026-09-29 10:41:38+00:00", "updated_at": "2026-09-29 10:49:13.540651+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "developer-tools", "agent-protocols", "artificial-intelligence"], "entities": ["EverMind", "Raven 0.2.0", "EverOS", "Shanda Group", "Nvidia", "Jensen Huang", "Open Agent Safety Platform", "Agent2Agent 1.0"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/raven-0-2-0-gives-developers-an-open-source-alternative-to-closed-ai-agent", "markdown": "https://wpnews.pro/news/raven-0-2-0-gives-developers-an-open-source-alternative-to-closed-ai-agent.md", "text": "https://wpnews.pro/news/raven-0-2-0-gives-developers-an-open-source-alternative-to-closed-ai-agent.txt", "jsonld": "https://wpnews.pro/news/raven-0-2-0-gives-developers-an-open-source-alternative-to-closed-ai-agent.jsonld"}}