[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-81040":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":14,"stars7d":15,"stars30d":16,"stars90d":13,"forks30d":13,"starsTrendScore":17,"compositeScore":18,"rankGlobal":8,"rankLanguage":8,"license":8,"archived":19,"fork":19,"defaultBranch":20,"hasWiki":21,"hasPages":19,"topics":22,"createdAt":8,"pushedAt":8,"updatedAt":23,"readmeContent":24,"aiSummary":25,"trendingCount":13,"starSnapshotCount":13,"syncStatus":26,"lastSyncTime":27,"discoverSource":28},81040,"kanban-video-pipeline","NousResearch\u002Fkanban-video-pipeline","NousResearch",null,"Python",34,7,28,0,1,4,6,3,2.71,false,"main",true,[],"2026-06-12 02:04:10","# Kanban Video Pipeline — Multi-Agent Creative Production\n\n> A recursive demonstration: the Hermes Agent Kanban system creates a video explaining itself.\n\nFour AI agent profiles collaborate autonomously — a director, cinematographer, renderer, and editor — communicating through structured handoffs, comments, and shared workspace files. One command kicks off the entire pipeline; you come back to a finished video.\n\n## What This Is\n\nA **Kanban-driven multi-agent video production pipeline** built on [Hermes Agent](https:\u002F\u002Fhermes-agent.nousresearch.com). You describe a video, the system decomposes it into a task graph, and specialized agent profiles execute each stage:\n\n```\nYou type one command\n    → Director decomposes the brief into 9 tasks\n        → Cinematographer designs visual language\n            → 7 Renderers execute in parallel (ASCII video + p5.js)\n                → Editor reviews clips + assembles final cut\n                    → output\u002Ffinal.mp4\n```\n\nThe entire pipeline auto-runs with zero human intervention. Every stage promotes automatically when its dependencies complete.\n\n## The Recursive Concept\n\nThis video is **about** Kanban — and it was **made by** Kanban. The first project demonstrates the system by having it create an explainer video about itself. Two-part structure:\n\n1. **Part 1** (captured with Screen Studio): You type the command, the dashboard lights up, tasks flow between columns\n2. **Part 2** (generated by the pipeline): The actual video the system produced — ASCII art + generative visuals explaining how Kanban works\n\n## The Team\n\n| Profile | Role | Communication |\n|---------|------|---------------|\n| **Director** | Vision, decomposition, final approval | Handoffs to cinematographer, comments on anything |\n| **Cinematographer** | Visual spec, quality gate | Specs to renderers, feedback via comments |\n| **Renderer** | Code execution (ascii-video \u002F p5.js) | Reads parent handoff + comments, writes clips |\n| **Editor** | Pacing review, assembly, final cut | Reviews with AI video analysis, assembles with ffmpeg |\n\n## Cross-Profile Communication\n\nAgents talk to each other through three built-in Kanban channels:\n\n- **Structured handoffs** — when a task completes with `summary` + `metadata`, every child task automatically receives it in context\n- **Comments** — any profile can write `kanban_comment(task_id=\u003Cany_task>)` on any task; the next worker reads the full thread\n- **Shared workspace** — all profiles read\u002Fwrite to the same project directory\n\n## Quick Start\n\n### Prerequisites\n\n- [Hermes Agent](https:\u002F\u002Fhermes-agent.nousresearch.com) installed\n- OpenRouter API key (for claude-opus-4.6)\n- Gateway running (`hermes gateway start`)\n\n### Setup\n\n```bash\nchmod +x setup.sh\n.\u002Fsetup.sh\n```\n\nThis creates 4 profiles (director, cinematographer, renderer, editor), writes their personalities, sets up the project structure, and initializes the Kanban board.\n\n### Run\n\n```bash\nhermes kanban create \"Make a video explaining how Hermes Kanban works\" \\\n    --assignee director \\\n    --workspace dir:$(pwd)\u002Fkanban-launch-video \\\n    --tenant video-pipeline \\\n    --priority 2 \\\n    --max-runtime 4h \\\n    --body \"Read brief.md. Decompose into: cinematographer → renderers → editor. All child tasks MUST use workspace_kind='dir' and workspace_path='$(pwd)\u002Fkanban-launch-video'. Pass tenant='video-pipeline' on every create.\"\n```\n\n### Monitor\n\n```bash\nhermes kanban watch --tenant video-pipeline    # live events\nhermes kanban list --tenant video-pipeline     # board state\nhermes dashboard                                # visual board (Kanban tab)\n```\n\n## Project Structure\n\n```\n├── README.md                     ← you are here\n├── ARCHITECTURE.md               ← system design, communication model\n├── RUN.md                        ← step-by-step execution guide\n├── brief.md                      ← the video specification\n├── setup.sh                      ← one-shot setup script\n├── taste\u002F\n│   ├── brand-guide.md            ← visual language rules\n│   └── emotional-dna.md          ← the creative soul document\n├── tools\u002F\n│   └── media_analyze.py          ← AI audio\u002Fvideo analysis CLI (Gemini via OpenRouter)\n└── render-scripts\u002F               ← what the agents actually wrote (reference)\n    ├── scene-01\u002Frender.py        ← ASCII video (Python + NumPy + ffmpeg)\n    ├── scene-02\u002Frender.py\n    ├── scene-03\u002Fsketch.html      ← p5.js generative (HTML + Puppeteer)\n    ├── scene-04\u002Frender.py\n    ├── scene-05\u002Fsketch.html\n    ├── scene-06\u002Frender.py\n    └── scene-07\u002Frender.py\n```\n\n## Output\n\nThe pipeline produces:\n- `output\u002Ffinal.mp4` — 1920×1080 @ 24fps, ~2 minutes, with audio\n- `output\u002Ffinal-noaudio.mp4` — silent version for re-scoring\n- Individual scene clips at `scenes\u002Fscene-NN\u002Fclip.mp4`\n\n## How It Works (Deep Dive)\n\nSee [ARCHITECTURE.md](.\u002FARCHITECTURE.md) for:\n- Full communication model diagram\n- Task graph with dependency links\n- Iteration loops (cinematographer ↔ renderer, editor ↔ renderer)\n- Renderer flexibility (swap ascii-video \u002F p5js \u002F remotion per task)\n- Media analysis tool usage\n\n## Timing\n\n| Phase | Duration | What Happens |\n|-------|----------|--------------|\n| Director | ~3 min | Reads brief, creates 9 tasks |\n| Cinematographer | ~8 min | Designs visual spec for all scenes |\n| Renderers (×7 parallel) | ~20 min | ASCII\u002Fp5.js rendering |\n| Editor | ~10 min | Reviews, assembles, muxes audio |\n| **Total** | **~45 min** | Fully autonomous |\n\n## Adapting This\n\nTo make your own video:\n1. Edit `brief.md` with your video spec\n2. Replace `taste\u002Fbrand-guide.md` with your visual language\n3. Drop your audio into the workspace\n4. Run `setup.sh` → fire the kanban command\n5. Come back to a video\n\nThe renderer profile has both `ascii-video` and `p5js` skills — use `--skill \u003Cname>` on individual tasks to control which renderer is used per scene.\n\n## License\n\nMIT\n","Kanban Video Pipeline 是一个基于 Kanban 的多智能体视频生产流水线，使用 Hermes Agent 构建。该项目的核心功能在于通过四个自主协作的AI角色——导演、摄影师、渲染师和剪辑师，将用户输入的视频描述自动分解为任务图，并执行每个阶段的任务，最终生成完整的视频文件。整个过程无需人工干预，各阶段在依赖项完成后自动推进。此项目适用于需要自动化创意视频制作的场景，特别是那些希望利用结构化工作流程来提高效率的团队或个人。",2,"2026-06-11 04:03:17","CREATED_QUERY"]