{"slug": "opus-5-5-is-good-at-explainer-videos", "title": "Opus 5.5 is good at explainer videos", "summary": "A new open-source project called shipvideo uses Anthropic's Claude Opus 5.5 to generate launch videos from a URL or product description, with a serverless agent on OpenComputer rendering each film in about four minutes for roughly 100k tokens. The agent runs on one OpenComputer serverless agent file with three tools — web_fetch, check_scene, and render_video — executing in a fresh microVM (Amazon Linux 2023 on arm64, 4 vCPU, 8 GB RAM, Node 22) that renders 1920x1080 at 30 fps with JPEG frames piped into libx264 at crf 18. The project, credited to an idea from Deedy's post on Opus 5.5 and instructional video, is deployable via `npx opencomputer template deploy https://github.com/diggerhq/shipvideo`.", "body_md": "# Ship a launch video.\n\nPaste a URL or describe the product. Opus 5.5 writes the film and a serverless agent renders it. About four minutes and roughly 100k tokens per video.\n\nexamples\n\n## Made by this page, untouched.\n\nEach one is a single run: a URL or a prompt in, an MP4 out. No edits.\n\nhow it runs\n\n## A serverless agent on OpenComputer. Yours in one click.\n\nThe whole product is one agent file, three tools, and this form. OpenComputer runs the agent, the microVM it renders in, the model gateway, and the session API the page polls.\n\n- Agent\n- One OpenComputer serverless agent, defined in TypeScript and deployed with `opencomputer deploy` . No framework, no queue, no server of ours.\n- Model\n- anthropic/claude-opus-5.5 through OpenComputer's model gateway. Roughly 90k input and 15k output tokens per film, most of it the HTML itself.\n- Runtime\n- Every job is one session in a fresh microVM: Amazon Linux 2023 on arm64, 4 vCPU, 8 GB RAM, Node 22. The first tool call installs Playwright's headless Chromium and a static ffmpeg (about a minute); the VM is thrown away after.\n- Tools\n- Three `defineTool` functions. web_fetch returns page text plus title, headings, the most used hex colors, and Google Fonts. check_scene loads the film and reports JS errors and the visible text at sample timestamps. render_video renders and uploads.\n- Rendering\n- No video model. The page's clocks (requestAnimationFrame, timers, Date, CSS and Web Animations) are replaced with a virtual clock, so every frame is a deterministic seek. 1920x1080 at 30 fps, JPEG frames piped into libx264, crf 18.\n- Storage\n- The agent holds no secrets. The form mints a Vercel Blob upload token scoped to one path for three hours, parks it in a per-job manifest, and the tool fetches it by job id. The finished MP4 is a public Blob URL.\n- Control plane\n- This page uses the same API the CLI does: create a session, send one turn, poll the event stream (tool.started, tool.completed, turn.completed) to show progress, and treat the MP4 appearing in Blob as done.\n\n``` js\n// opencomputer/agents/director/agent.ts\nimport { useInput, useModel, useTool } from \"@opencomputer/agent\";\nimport { checkScene, renderVideo } from \"./tools/scene.js\";\nimport { webFetch } from \"./tools/web.js\";\n\nexport default function Agent() {\n  const input = useInput();           // the JOB block from the form\n  useModel(\"anthropic/claude-opus-5.5\");\n  useTool(webFetch);                  // read the product's site\n  useTool(checkScene);                // load the HTML, report errors + visible text\n  useTool(renderVideo);               // headless Chromium → ffmpeg → Blob\n  return `You are a motion designer who writes code. ...`;\n}\n```\n\nDeploy this agent\n\nOne click. Free account, the agent lands in your project with its tools and prompt.\n\nClone the repo\n\ndiggerhq/shipvideo: the agent, the renderer, and this web app.\n\nBuild your own agent\n\nThe quickstart: a TypeScript file, a deploy, a session. Ten minutes.\n\n`npx opencomputer template deploy https://github.com/diggerhq/shipvideo`\nEverything above is in the repo, and [one click deploys it to your account](https://app.opencomputer.dev/new?repository-url=https%3A%2F%2Fgithub.com%2Fdiggerhq%2Fshipvideo). The idea comes from [Deedy's post](https://x.com/deedydas/status/2102787937482252537) on Opus 5.5 and instructional video: the model writes the film as code, and code renders the same every time.", "url": "https://wpnews.pro/news/opus-5-5-is-good-at-explainer-videos", "canonical_source": "https://launchvideo.io", "published_at": "2026-09-24 20:28:47+00:00", "updated_at": "2026-09-24 20:59:31.754291+00:00", "lang": "en", "topics": ["generative-ai", "large-language-models", "ai-agents", "ai-tools", "developer-tools"], "entities": ["Anthropic", "Claude Opus 5.5", "OpenComputer", "shipvideo", "diggerhq", "Deedy", "Playwright", "Vercel Blob"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/opus-5-5-is-good-at-explainer-videos", "markdown": "https://wpnews.pro/news/opus-5-5-is-good-at-explainer-videos.md", "text": "https://wpnews.pro/news/opus-5-5-is-good-at-explainer-videos.txt", "jsonld": "https://wpnews.pro/news/opus-5-5-is-good-at-explainer-videos.jsonld"}}