[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-93029":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":13,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":15,"stars7d":15,"stars30d":13,"stars90d":15,"forks30d":15,"starsTrendScore":15,"compositeScore":16,"rankGlobal":10,"rankLanguage":10,"license":17,"archived":18,"fork":18,"defaultBranch":19,"hasWiki":20,"hasPages":20,"topics":21,"createdAt":10,"pushedAt":10,"updatedAt":22,"readmeContent":23,"aiSummary":24,"trendingCount":15,"starSnapshotCount":15,"syncStatus":25,"lastSyncTime":26,"discoverSource":27},93029,"roger","rogerai-fyi\u002Froger","rogerai-fyi","RogerAI - a two-way radio for GPUs: a marketplace for self-hosted LLMs (CLI + broker + site).","https:\u002F\u002Frogerai.fyi",null,"Go",193,1,190,0,41,"Other",false,"main",true,[],"2026-07-22 04:02:07","# RogerAI\n\n[![coverage](https:\u002F\u002Frogerai-fyi.github.io\u002Froger\u002Fcoverage\u002Fbadge.svg)](https:\u002F\u002Frogerai-fyi.github.io\u002Froger\u002Fcoverage\u002Fpackages.html)\n\n```\n             ·   ˙   ·\n           ˙ ((  •  )) ˙                   ┌──────────────┐\n              \\(   )\u002F                      │ ((•)) ON AIR │\n               │ R │         ▟█▙           └──────────────┘\n   ┌───────────┴───┴──────────█───────────────────────────┐\n   │  ◉ ch 01 · on band gpt-oss-20b · 42 tok\u002Fs   ▂ ▄ ▆ █  │\n   ╰┬────────────────────────────────────────────────────┬╯\n    ▔                                                    ▔\n\n        █▀▀█ █▀▀█ █▀▀▀ █▀▀▀ █▀▀█     █▀▀█ ▀█▀  ▄\n        █▀▀▄ █  █ █ ▀█ █▀▀  █▀▀▄  ▄  █▀▀█  █   █\n        ▀  ▀ ▀▀▀▀ ▀▀▀▀ ▀▀▀▀ ▀  ▀     ▀  ▀ ▀▀▀  ▀\n             borrow a GPU, pay by the token · roger that.\n```\n\n**A two-way radio for GPUs.** RogerAI is a marketplace for crowd-sourced, self-hosted LLMs:\npeople run open models on their own GPUs and go \"on air\"; you tune in and pay per token. Every\ntoken carries a **model-lineage record** - a receipt signed by the provider and counter-signed\nby the broker, tracing the response back to the model that produced it. Owners monetize idle\nhardware; users get cheap, diverse access.\n\n```\ncurl -fsSL https:\u002F\u002Frogerai.fyi\u002Finstall.sh | sh\n```\n\nOr via **Homebrew** (macOS + Linux). Homebrew 6+ makes you trust a third-party tap once, so\nit's a single line:\n\n```\nbrew trust rogerai-fyi\u002Fhomebrew-tap && brew install rogerai-fyi\u002Fhomebrew-tap\u002Froger\n```\n\nThen just run `roger` for the interactive radio (browse stations, tune in, test, copy the\nendpoint; `rogerai` is the legacy alias). Already have Go?\n`go install github.com\u002Frogerai-fyi\u002Froger\u002Fcmd\u002Frogerai@latest`.\n\n## Use it\n\n```\nroger                         # interactive TUI: browse → connect → test → copy endpoint\nroger search                  # list models (cheapest now first; shows tok\u002Fs, ◆ confidential, FREE)\nroger use \u003Cmodel>             # open a local OpenAI-compatible endpoint that relays via the broker\nroger voices                  # browse voices on air (TTS\u002FSTT stations, sample previews)\nroger say --voice \u003Cv> \"hi\"    # text-to-speech through a voice station, played locally\nroger balance                 # wallet credits\nroger topup 10                # buy credits (Stripe)\nroger remote                  # your private BASE STATION: continue an agent session from anywhere\n```\n\nInside the agent (`[0] AGENT`), `\u002Fremote-control` puts the live session on your private **Base\nStation** so you can pick it up from another terminal (`roger remote attach \u003Ccode>`), the web\nconsole, or the app - private to your account, one-time link codes for a phone. Tools keep\nrunning on the host and still ask before anything mutating.\n\n`roger use` (or the TUI's \"tune in\") exposes `http:\u002F\u002F127.0.0.1:4141\u002Fv1` with an API key - point\nany OpenAI-compatible tool at it.\n\n## Share your GPU (become a provider)\n\nThe node **dials OUT** and long-polls the broker for jobs - no inbound ports, no tunnel\ndependency. Behind any NAT, one command:\n\n```\nroger share                                    # auto-detects your local model, starts earning\n# options: --price-in\u002F--price-out, --free-window 03:00-03:30, --schedule '\u003Ctime-of-use JSON>',\n#          --confidential (TEE-attested), --upstream \u003Cyour OpenAI endpoint>\n```\n\n**Bring your own runner.** The node relays to any **OpenAI-compatible** endpoint, so it works with\nwhatever you already host - **Ollama, llama.cpp, LM Studio, vLLM, Osaurus, Jan, LocalAI, TGI,\nSGLang, KoboldCpp, text-generation-webui** - and anything else that speaks the OpenAI API (point it\nwith `--upstream`). Same on the other side: `roger use` gives you an OpenAI-compatible endpoint, so\nany tool that talks to OpenAI talks to RogerAI.\n\n## How it works\n\n```\nrogerai ──► broker (broker.rogerai.fyi) ──► your node ──► your local model\n discover     registry · wallet · relay      dials out      (Ollama\u002Fllama.cpp\u002FLM Studio\u002FvLLM\u002FOsaurus\u002F…)\n use\u002Ftopup    match · meter · co-sign         serve+sign\n```\n\nThe broker is the only public component and is **content-blind** (it stores token counts and\nsigned receipts, never prompts). It's an OpenAI-compatible relay - see the served spec at\n`\u002Fopenapi.yaml`.\n\n- **Per-token pricing** with a 24h price-lock, **free** and **time-of-use** windows.\n- **Voice bands** - TTS\u002FSTT stations alongside chat: share a Kokoro-style speech server with\n  `roger share`, browse with `roger voices`, attributed on air as `@station\u002Fslug`.\n- **Lineage receipts** - hash-chained, dual-signed (`internal\u002Fprotocol`).\n- **Privacy** - identity pseudonymized to providers; a confidential (TEE) tier for sensitive work.\n- **Routing constraints** - price, measured throughput (tok\u002Fs), confidential-only.\n\n## Run it locally\n\n```\nmake build\nmake demo     # broker + a node + a request, end to end (needs a local OpenAI endpoint)\ngo test .\u002F...\n```\n\n## Docs\n\n- [BROKER-SPEC.md](BROKER-SPEC.md) - the open broker spec (anyone can self-host\u002Ffederate)\n- `cmd\u002Frogerai-broker\u002Fopenapi.yaml` - OpenAPI 3.1 (also served at `\u002Fopenapi.yaml`)\n- [VERIFICATION.md](VERIFICATION.md) · [PRIVACY.md](PRIVACY.md) · [DEPLOY.md](DEPLOY.md) · [STRIPE.md](STRIPE.md)\n\n## License\n\n[PolyForm Perimeter 1.0.0](LICENSE) - source-available, free for any non-competing use. You can\nself-host your own broker for your own community; you can't run it as a competing commercial\nmarketplace service. See the LICENSE for the exact grant.\n","RogerAI 是一个面向自托管大语言模型的去中心化市场平台，支持用户按需调用他人闲置GPU上运行的开源LLM，并按实际生成token计费。核心功能包括：基于CLI的交互式TUI终端、OpenAI兼容API代理、模型谱系追踪（带签名的token级溯源凭证）、本地语音合成\u002F识别接入、以及无需端口暴露的反向连接节点部署机制。技术上采用Go实现，强调隐私与可验证性，适用于个人开发者快速试用多样化开源模型、研究者低成本获取异构模型服务，以及硬件持有者将空闲GPU资源变现的场景。",2,"2026-07-11 02:30:42","CREATED_QUERY"]