[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-75833":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":8,"htmlUrl":8,"language":9,"languages":8,"totalLinesOfCode":8,"stars":10,"forks":11,"watchers":12,"openIssues":13,"contributorsCount":13,"subscribersCount":13,"size":13,"stars1d":13,"stars7d":14,"stars30d":15,"stars90d":13,"forks30d":13,"starsTrendScore":16,"compositeScore":17,"rankGlobal":8,"rankLanguage":8,"license":8,"archived":18,"fork":18,"defaultBranch":19,"hasWiki":20,"hasPages":18,"topics":21,"createdAt":8,"pushedAt":8,"updatedAt":22,"readmeContent":23,"aiSummary":24,"trendingCount":13,"starSnapshotCount":13,"syncStatus":16,"lastSyncTime":25,"discoverSource":26},75833,"FAROS","OpenNSWM-Lab\u002FFAROS","OpenNSWM-Lab",null,"Python",1026,171,127,0,163,915,2,19.71,false,"main",true,[],"2026-06-12 02:03:36","\u003Cp align=\"center\">\n  \u003Ch1 align=\"center\">FAROS\u003C\u002Fh1>\n  \u003Cp align=\"center\">\u003Cb>Foundation AutoResearch Operating System\u003C\u002Fb>\u003C\u002Fp>\n  \u003Cp align=\"center\">\u003Ci>Blueprint-driven AutoResearch runtime for the LLM domain today, extensible research workflows tomorrow.\u003C\u002Fi>\u003C\u002Fp>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FRuntime-FAROS--LLM-0f766e?style=for-the-badge\" alt=\"FAROS-LLM\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPhase-1.1.0--rc1-7c3aed?style=for-the-badge\" alt=\"1.1.0-rc1\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPython-3.11%2B-3776AB?style=for-the-badge&logo=python&logoColor=white\" alt=\"Python 3.11+\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FNode.js-18%2B-339933?style=for-the-badge&logo=node.js&logoColor=white\" alt=\"Node 18+\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FTests-10%20passed-16a34a?style=for-the-badge&logo=pytest&logoColor=white\" alt=\"10 passed\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDomain-LLM%20Research-d97706?style=for-the-badge\" alt=\"LLM Research\">\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"#-release-scope\">Release Scope\u003C\u002Fa> ·\n  \u003Ca href=\"#-why-faros\">Why FAROS\u003C\u002Fa> ·\n  \u003Ca href=\"#-current-workflow\">Workflow\u003C\u002Fa> ·\n  \u003Ca href=\"#-architecture\">Architecture\u003C\u002Fa> ·\n  \u003Ca href=\"#-quick-start\">Quick Start\u003C\u002Fa> ·\n  \u003Ca href=\"#-faros-api\">API\u003C\u002Fa> ·\n  \u003Ca href=\"#-important-todo\">TODO\u003C\u002Fa>\n\u003C\u002Fp>\n\n\n\n---\n\n> [!IMPORTANT]\n> FAROS is not a single hardcoded AI scientist agent. It is a research workflow runtime built around **Blueprints**, **Capabilities**, **Profiles**, and **Providers**.\n>\n> This release ships the first runnable baseline: **FAROS-LLM**.\n\n## ✨ Tagline\n\n\u003Cp align=\"center\">\n  \u003Cb>\u003Ci>Define a research workflow. Bind a profile. Run an AutoResearch system.\u003C\u002Fi>\u003C\u002Fb>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ccode>idea -> experiment -> paper -> review\u003C\u002Fcode>\n\u003C\u002Fp>\n\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\".\u002FFAROS.png\" alt=\"FAROS architecture overview\" width=\"88%\">\n\u003C\u002Fp>\n\n---\n\n## 📦 Release Scope\n\nThis repository is the current release candidate for the **LLM-domain FAROS baseline**.\nIt is already a runnable AutoResearch runtime, but it is not yet the final cross-domain platform vision.\n\n\u003Ctable>\n\u003Ctr>\n\u003Ctd width=\"50%\" valign=\"top\">\n\n### Included\n\n- FAROS runtime under `backend\u002Fapp\u002Ffaros`\n- Blueprint loading and profile loading\n- Capability and provider registries\n- File-backed run, event, artifact, and memory persistence\n- First blueprint: `ml_paper`\n- First profile: `faros_llm`\n- Complete LLM workflow: `idea -> experiment -> paper -> review`\n- Existing module-native APIs for `idea`, `code`, `paper`, `review`, `platform`\n- Venue-aware LaTeX paper generation\n\n\u003C\u002Ftd>\n\u003Ctd width=\"50%\" valign=\"top\">\n\n### Not Yet Included\n\n- full DAG scheduling and parallel orchestration\n- generalized non-LLM provider ecosystem\n- full experiment execution and evaluation loop\n- FAROS frontend console\n- DB-backed FAROS runtime metadata\n- mature cross-domain blueprint library\n\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n---\n\n## 🤔 Why FAROS\n\nMost \"AI scientist\" systems are built as one fixed application with one workflow and one set of assumptions.\nFAROS takes a different approach: treat research automation as a **runtime problem**, not a single-agent prompt stack.\n\n\u003Ctable>\n\u003Ctr>\u003Cth width=\"24%\">Layer\u003C\u002Fth>\u003Cth>Responsibility\u003C\u002Fth>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Cb>Blueprint\u003C\u002Fb>\u003C\u002Ftd>\u003Ctd>Defines the workflow graph, constraints, output contract, and validation requirements.\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Cb>Capability\u003C\u002Fb>\u003C\u002Ftd>\u003Ctd>Implements one executable research step such as idea refinement, experiment provisioning, paper drafting, or reviewer simulation.\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Cb>Profile\u003C\u002Fb>\u003C\u002Ftd>\u003Ctd>Binds a blueprint to a concrete execution strategy.\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Cb>Provider\u003C\u002Fb>\u003C\u002Ftd>\u003Ctd>Supplies the actual engine behind a capability, such as LLM, tool, API, or human review.\u003C\u002Ftd>\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n> [!NOTE]\n> In FAROS, LLM is only one provider class. This release ships `FAROS-LLM`, but the runtime is being shaped so future domains can plug in other providers without rewriting the core orchestration layer.\n\n---\n\n## 🚀 What Makes This Release Different\n\n\u003Ctable>\n\u003Ctr>\u003Cth width=\"24%\">Principle\u003C\u002Fth>\u003Cth>How This Release Applies It\u003C\u002Fth>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Cb>Keep What Works\u003C\u002Fb>\u003C\u002Ftd>\u003Ctd>The current `idea`, `code`, `paper`, `review`, and `platform` modules are reused through FAROS capability adapters instead of being replaced by a second parallel application.\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Cb>Add a Runtime Boundary\u003C\u002Fb>\u003C\u002Ftd>\u003Ctd>New orchestration logic lives under `backend\u002Fapp\u002Ffaros`, giving memory, verification, profiles, and providers a stable place to evolve.\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Cb>Finish One Domain First\u003C\u002Fb>\u003C\u002Ftd>\u003Ctd>The first complete chain is the LLM research domain. Cross-domain abstraction comes after the first workflow is coherent and runnable.\u003C\u002Ftd>\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n---\n\n## 🔄 Current Workflow\n\nThe first FAROS blueprint is `ml_paper`.\n\n\u003Ctable>\n\u003Ctr>\u003Cth width=\"18%\">Stage\u003C\u002Fth>\u003Cth width=\"22%\">Capability\u003C\u002Fth>\u003Cth>Output\u003C\u002Fth>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>1\u003C\u002Ftd>\u003Ctd>\u003Ccode>idea_refinement\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Idea session, ranked candidates, selected candidate\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>2\u003C\u002Ftd>\u003Ctd>\u003Ccode>experiment\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Code project scaffold and experiment record for the LLM domain\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>3\u003C\u002Ftd>\u003Ctd>\u003Ccode>paper_drafting\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Venue-aware LaTeX project, PDF, and paper artifacts\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>4\u003C\u002Ftd>\u003Ctd>\u003Ccode>reviewer_simulation\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Structured review plus actionable follow-up items\u003C\u002Ftd>\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Current Artifact Surface\n\n\u003Ctable>\n\u003Ctr>\u003Cth width=\"22%\">Artifact Type\u003C\u002Fth>\u003Cth>Description\u003C\u002Fth>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>idea_session\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Idea generation session with ranked candidate outputs\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>code_project\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Provisioned research code workspace for the experiment stage\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>experiment_record\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Experiment metadata record for the LLM workflow\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>latex_project\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Paper source bundle with venue-aware LaTeX assets\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>paper_pdf\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Compiled paper PDF or fallback rendered PDF\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>review_report\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Structured review with action items\u003C\u002Ftd>\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n---\n\n## 🏗️ Architecture\n\n```text\nbackend\u002Fapp\u002F\n  faros\u002F\n    api\u002F\n    blueprints\u002F\n    capabilities\u002F\n    loaders\u002F\n    memory\u002F\n    models\u002F\n    profiles\u002F\n    providers\u002F\n    registry\u002F\n    runtime\u002F\n    verification\u002F\n  modules\u002F\n    idea\u002F\n    code\u002F\n    paper\u002F\n    review\u002F\n    platform\u002F\n```\n\n### Runtime Layers\n\n\u003Ctable>\n\u003Ctr>\u003Cth width=\"24%\">Area\u003C\u002Fth>\u003Cth>Role\u003C\u002Fth>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Cb>FAROS Runtime\u003C\u002Fb>\u003C\u002Ftd>\u003Ctd>Blueprint loading, capability registry, profile binding, orchestrated execution, event logging, artifact persistence, research memory, and baseline verification\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Cb>Domain Modules\u003C\u002Fb>\u003C\u002Ftd>\u003Ctd>Reusable implementation surfaces for `idea`, `code`, `paper`, `review`, and `platform`\u003C\u002Ftd>\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Execution Model\n\n```mermaid\nflowchart LR\n    B[Blueprint] --> O[Orchestrator]\n    P[Profile] --> O\n    R[Provider Registry] --> O\n    C[Capability Registry] --> O\n    O --> I[idea_refinement]\n    I --> E[experiment]\n    E --> W[paper_drafting]\n    W --> V[reviewer_simulation]\n    O --> M[Research Memory]\n    O --> A[Artifacts]\n    O --> Q[Verification]\n```\n\n---\n\n## 🗂️ Repository Layout\n\n```text\ngithub-v1\u002F\n  backend\u002F\n    app\u002F\n      faros\u002F\n      modules\u002F\n      llm\u002F\n      db\u002F\n      storage\u002F\n    templates\u002Flatex\u002F\n    tests\u002F\n  frontend\u002F\n    src\u002F\n  docs\u002F\n    DEVELOPER_GUIDE.md\n    FAROS_TODO.md\n```\n\n---\n\n## ⚙️ Runtime Requirements\n\n- Python `3.11+` or `3.12`\n- Node.js `18+`\n- `latexmk` and `pdflatex` for venue-style PDF compilation\n- a configured LLM provider for real execution\n\n> [!TIP]\n> The development environment used during this release cycle has been the conda environment `aist`.\n\n---\n\n## 🚀 Quick Start\n\n### Backend\n\n```bash\ncd backend\npip install -r requirements.txt\nuvicorn app.main:app --host 127.0.0.1 --port 8005 --reload\n```\n\n### Frontend\n\n```bash\ncd frontend\nnpm install\nnpm run dev\n```\n\n### Useful Endpoints\n\n\u003Ctable>\n\u003Ctr>\u003Cth width=\"34%\">Endpoint\u003C\u002Fth>\u003Cth>Purpose\u003C\u002Fth>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Fsystem\u002Fhealth\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Basic backend health\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Fsystem\u002Fversion\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Release metadata\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Fdocs\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>OpenAPI docs\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Ffaros\u002Fhealth\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>FAROS runtime health\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Ffaros\u002Fblueprints\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Available FAROS blueprints\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Ffaros\u002Fprofiles\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Available FAROS profiles\u003C\u002Ftd>\u003C\u002Ftr>\n\u003C\u002Ftable>\n\nIf needed, set `VITE_API_BASE_URL` for the frontend.\n\n---\n\n## 🔐 Provider Configuration\n\nThe backend supports multiple providers, including `minimax`.\n\nConfiguration is loaded from:\n1. environment variables defined in `backend\u002Fapp\u002Fcore\u002Fsettings.py`\n2. runtime settings persisted to `backend\u002Fdata\u002Fprovider_config.json`\n\nDo not commit real API keys.\n\n---\n\n## 🔌 FAROS API\n\n\u003Ctable>\n\u003Ctr>\u003Cth width=\"38%\">Endpoint\u003C\u002Fth>\u003Cth>Purpose\u003C\u002Fth>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Ffaros\u002Fhealth\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Runtime health and asset counts\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Ffaros\u002Fblueprints\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>List available blueprints\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Ffaros\u002Fprofiles\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>List available profiles\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Ffaros\u002Fcapabilities\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>List registered capabilities and providers\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Ffaros\u002Fruns\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>List FAROS runs\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>POST \u002Fapi\u002Ffaros\u002Fruns\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Create a FAROS run\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Ffaros\u002Fruns\u002F{run_id}\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Inspect one FAROS run\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Ffaros\u002Fruns\u002F{run_id}\u002Fevents\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Inspect run events\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>GET \u002Fapi\u002Ffaros\u002Fruns\u002F{run_id}\u002Fartifacts\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Inspect run artifacts\u003C\u002Ftd>\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Example: Plan-Only Run\n\n```bash\ncurl -X POST http:\u002F\u002F127.0.0.1:8005\u002Fapi\u002Ffaros\u002Fruns   -H 'Content-Type: application\u002Fjson'   -d '{\n    \"blueprintId\": \"ml_paper\",\n    \"profileId\": \"faros_llm\",\n    \"executionMode\": \"plan\",\n    \"inputs\": {\n      \"seedQuery\": \"Improve CPU efficiency in LLM workflows\",\n      \"paperType\": \"system\",\n      \"targetVenue\": \"generic\"\n    }\n  }'\n```\n\n---\n\n## 📝 Paper Generation\n\nPaper generation in this release uses real venue-aware LaTeX template assets.\n\n\u003Ctable>\n\u003Ctr>\u003Cth width=\"22%\">Template\u003C\u002Fth>\u003Cth>Description\u003C\u002Fth>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>icml\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>ICML-style LaTeX template path\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>neurips\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>NeurIPS-style LaTeX template path\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>iclr\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>ICLR-style LaTeX template path\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>acl\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>ACL-style LaTeX template path\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>generic\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Fallback generic template\u003C\u002Ftd>\u003C\u002Ftr>\n\u003C\u002Ftable>\n\nCompilation prefers `latexmk`.\nIf LaTeX compilation fails, the backend falls back to simplified PDF rendering so the workflow still yields a previewable artifact.\n\n---\n\n## ✅ Verification\n\n### Release Checks\n\n```bash\nbash scripts\u002Fcheck_release.sh\n```\n\n```bash\nbash backend\u002Fscripts\u002Fcheck_backend_release.sh\n```\n\n```bash\nbash frontend\u002Fscripts\u002Fcheck_frontend_release.sh\n```\n\n### Current Validation State\n\n- `github-v1\u002Fbackend\u002Ftests` in `aist`: `10 passed`\n- FAROS runtime routes mounted\n- plan-mode FAROS run creation verified\n- LLM-domain FAROS workflow skeleton wired through `idea -> experiment -> paper -> review`\n\n---\n\n## 🧱 Stable Surface In This Release\n\nThese parts should be treated as the release baseline:\n\n\u003Ctable>\n\u003Ctr>\u003Cth width=\"30%\">Area\u003C\u002Fth>\u003Cth>Stability Statement\u003C\u002Fth>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>\u003Ccode>backend\u002Fapp\u002Ffaros\u002F*\u003C\u002Fcode>\u003C\u002Ftd>\u003Ctd>Primary runtime surface for future FAROS work\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>Blueprint\u002FProfile loading\u003C\u002Ftd>\u003Ctd>Stable release baseline\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>FAROS metadata API\u003C\u002Ftd>\u003Ctd>Stable release baseline\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>Plan-mode FAROS run creation\u003C\u002Ftd>\u003Ctd>Stable release baseline\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>Provider settings path\u003C\u002Ftd>\u003Ctd>Stable release baseline\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>Paper generation path\u003C\u002Ftd>\u003Ctd>Stable release baseline\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>Review generation path\u003C\u002Ftd>\u003Ctd>Stable release baseline\u003C\u002Ftd>\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n---\n\n## 📌 Important TODO\n\nThe most important next steps after this release are:\n- replace the current `experiment` scaffold with true code synthesis and execution for the LLM domain\n- connect experiment outputs to metrics ingestion, figure generation, and run tracking\n- replace linear graph execution with a real DAG runtime\n- add stronger verification beyond required-key checks\n- add a dedicated FAROS frontend console\n- add provider inheritance policies instead of hardcoded profile defaults\n\nSee [docs\u002FFAROS_TODO.md](docs\u002FFAROS_TODO.md) for the detailed backlog.\n\n---\n\n## 🛠️ Development Notes\n\nUse [docs\u002FDEVELOPER_GUIDE.md](docs\u002FDEVELOPER_GUIDE.md) for module ownership, extension boundaries, and development conventions.\n\nCurrent working rule:\n- extend FAROS under `backend\u002Fapp\u002Ffaros`\n- keep domain-specific logic inside `backend\u002Fapp\u002Fmodules\u002F*`\n- avoid adding new business logic to legacy compatibility paths unless required for release stability\n\n---\n\n## 📍 Project Status\n\nThis repository is the first FAROS release candidate.\nIt is already usable as a runtime baseline for LLM-domain AutoResearch workflows, but it is still the beginning of the platform transition rather than the end state.\n\n\n\n# GitHub Stars\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fstar-history.com\u002F#OpenNSWM-Lab\u002FFAROS&Date\">\n    \u003Cimg src=\"https:\u002F\u002Fapi.star-history.com\u002Fsvg?repos=OpenNSWM-Lab\u002FFAROS&type=Date\" alt=\"Star History Chart\" width=\"100%\"\u002F>\n  \u003C\u002Fa>\n\u003C\u002Fp>\n  \u003Csub>Built with care for the research community.\u003C\u002Fsub>\n\u003C\u002Fp>\n","FAROS 是一个基于蓝图驱动的自动化研究运行时系统，专为大规模语言模型（LLM）领域设计。其核心功能包括蓝图加载、配置文件绑定以及能力与提供者注册机制，支持从想法生成到论文撰写及审阅的完整研究流程。技术上，FAROS 使用 Python 3.11+ 和 Node.js 18+ 构建，并通过文件系统实现运行记录、事件和工件的持久化存储。目前版本已具备完整的 LLM 研究工作流支持，但尚未包含全面的数据流图调度、非 LLM 提供者生态系统等高级特性。该系统适用于需要构建灵活可扩展的研究流程，特别是专注于自然语言处理领域的科研团队或个人开发者。","2026-06-11 03:53:27","CREATED_QUERY"]