diff --git a/README.md b/README.md
index 4600405..405cb68 100644
--- a/README.md
+++ b/README.md
@@ -7,7 +7,7 @@
 
 English | [简体中文](./README.zh-CN.md)
 
-**A visual language for coding agents to explain ideas.**
+**Evidence-backed explainer videos for coding agents.**
 
 Seqvio gives coding agents a capture-to-explanation path for turning real terminal/browser activity and technical ideas into clear narrated videos. Human-readable `EDITORIAL.md` and `VISUAL-DESIGN.md` artifacts make content choices and visual direction reviewable before the executable `ExplainerDocument` IR binds spoken cues to visual actions with `ExplanationBeat`s.
 
@@ -15,15 +15,17 @@ Seqvio gives coding agents a capture-to-explanation path for turning real termin
 
 ## Demo
 
-Pre-rendered overview videos with CosyVoice narration, covering all three visual style packages (`@seqvio/whiteboard`, `@seqvio/scatterbrain`, `@seqvio/product-demo`). Source compositions live in [`examples/compositions/`](./examples/compositions/).
+The current 720p product demo shows the reviewable plan, executable
+`ExplainerDocument`, phrase-level `ExplanationBeat`, QA, and local rendering
+path through a native-module CI diagnosis.
 
-**English overview** — [`seqvio-overview-en.tsx`](./examples/compositions/seqvio-overview-en.tsx)
+**[Watch the current narrated demo](./docs/assets/videos/seqvio-product-hunt-en.mp4)**
+— source: [`seqvio-product-hunt-en.tsx`](./examples/compositions/seqvio-product-hunt-en.tsx)
 
-https://github.com/user-attachments/assets/83687d9c-63f0-4544-a67a-8f6eacc19928
-
-**中文介绍** — [`seqvio-overview-zh.tsx`](./examples/compositions/seqvio-overview-zh.tsx)
-
-https://github.com/user-attachments/assets/3ce605bc-7ad1-449b-a67c-5d8368f5398b
+Localized overview compositions remain available under
+[`examples/compositions/`](./examples/compositions/), but their previously
+published videos predate the current explanation contract and are no longer the
+primary product demo.
 
 ## Quick Start
 
@@ -61,7 +63,10 @@ npm install -g @seqvio/renderer
 seqvio-render --help
 ```
 
-Published packages: `@seqvio/core`, `@seqvio/whiteboard`, `@seqvio/scatterbrain`, `@seqvio/product-demo`, `@seqvio/technical`, and `@seqvio/renderer`.
+Public packages: `@seqvio/core`, `@seqvio/whiteboard`, `@seqvio/scatterbrain`,
+`@seqvio/product-demo`, `@seqvio/technical`, and `@seqvio/renderer`.
+Experimental capture packages: `@seqvio/capture`, `@seqvio/browser-recorder`,
+and `@seqvio/terminal-narrator`.
 
 Install optional style/component packages when a composition imports them directly:
 
@@ -97,7 +102,11 @@ After steps 1 and 2, try a prompt like:
 
 > Using `/seqvio`, first create and review an editorial plan and visual design brief, then compile them into a 4-scene Chinese ExplainerDocument with phrase-anchored visual Beats, run QA, and render the final MP4.
 
-The skill guides the agent through: pick an example composition, edit TSX, extract narration metadata, synthesize audio, and run `seqvio-render`.
+The recommended path reviews `EDITORIAL.md` and `VISUAL-DESIGN.md`, compiles an
+`ExplainerDocument`, resolves narration timing, runs `seqvio-qa`, and renders
+the MP4. Hand-authored TSX remains the lower-level path for deliberate visual
+control. Experimental terminal/browser adapters can supply observed scenes to
+the same IR; they are not required for authored explainers.
 
 Supported agents include Cursor, Claude Code, Codex, Gemini CLI, and other coding agents that support skills.
 
@@ -189,7 +198,7 @@ The skill lives in [`skills/seqvio/SKILL.md`](./skills/seqvio/SKILL.md) with sup
 | [`audio-workflow.md`](./skills/seqvio/references/audio-workflow.md) | Extract, synthesize, and mux narration |
 | [`render-workflow.md`](./skills/seqvio/references/render-workflow.md) | Build, render, and smoke-test commands |
 | [`production-techniques.md`](./skills/seqvio/references/production-techniques.md) | Voice-first timing, reference-style analysis, and visual QA rules |
-| [`planning-workflow.md`](./skills/seqvio/references/planning-workflow.md) | Storyboard IR planning and agent handoff |
+| [`planning-workflow.md`](./skills/seqvio/references/planning-workflow.md) | Editorial/visual planning and agent handoff |
 
 Install the skill (see [Quick Start](#quick-start)):
 
@@ -223,8 +232,8 @@ Seqvio is the visual language for coding agents that need to explain, not merely
 - Browser scene support with recorded video, cursor/focus/click metadata, exact action clocks, and time-mapped media seeking
 - `ExplanationBeat` cues, exact phrase anchors, visual actions, capture evidence, post-TTS `outputFrame`s, and semantic `sceneTimings[].timeMap`
 - `@seqvio/core` scene and transition primitives: `VideoComposition`, `Scene`, `Transition`
-- ExplainerDocument and retained Storyboard IR schema, validation, pacing,
-  and TSX compilation helpers
+- ExplainerDocument as the canonical IR, with retained Storyboard IR compatibility
+  for whiteboard-only input
 - `seqvio-render` CLI for TSX-to-MP4 rendering
 - `seqvio-audio` CLI for audio/caption manifest extraction and TTS synthesis
 - `seqvio-qa` CLI with baseline/capture profiles, stable audio/temporal/media diagnostics, configurable warning promotion, and key-frame visual checks
@@ -295,6 +304,7 @@ Voiceover is muxed automatically from the manifest. **Do not** add `--burnCaptio
 | [`@seqvio/technical`](./packages/technical) | Technical explainer runtime: code walkthroughs, architecture diagrams, terminal demos, annotations, and bundled fonts |
 | [`@seqvio/terminal-narrator`](./packages/terminal-narrator) | Pre-stable node-pty/xterm capture → IR/ExplanationBeat → optional narrated MP4 |
 | [`@seqvio/browser-recorder`](./packages/browser-recorder) | Pre-stable Chromium action capture with exact action timing → IR/ExplanationBeat |
+| [`@seqvio/capture`](./packages/capture) | Shared experimental capture session and artifact contracts |
 | [`@seqvio/renderer`](./packages/renderer) | TSX bundler plus `seqvio-render` and `seqvio-audio` CLIs |
 
 ## Documentation
@@ -310,6 +320,8 @@ Recommended reading:
 - [`examples/compositions/README.md`](./examples/compositions/README.md) — example catalog and conventions
 - [`skills/seqvio/SKILL.md`](./skills/seqvio/SKILL.md) — agent production loop
 - [`skills/seqvio/references/production-techniques.md`](./skills/seqvio/references/production-techniques.md) — narrated explainer production rules and QA checklist
+- [`docs/marketing/POSITIONING.md`](./docs/marketing/POSITIONING.md) — current product positioning and capability boundary
+- [`docs/marketing/FEATURE-STATUS.md`](./docs/marketing/FEATURE-STATUS.md) — public versus experimental wording
 
 If documentation conflicts with code, treat the code and [`docs/COMPOSITION-AUTHORING.md`](./docs/COMPOSITION-AUTHORING.md) as the source of truth.
 
diff --git a/README.zh-CN.md b/README.zh-CN.md
index 618334e..41eaf73 100644
--- a/README.zh-CN.md
+++ b/README.zh-CN.md
@@ -6,7 +6,7 @@
 
 [English](./README.md) | 简体中文
 
-**让 coding agent 拥有解释想法的视觉语言。**
+**让 coding agent 把真实技术工作变成有证据的讲解视频。**
 
 Seqvio 为 coding agent 提供从真实系统捕获到讲解视频的完整路径。人类可读的 `EDITORIAL.md` 与 `VISUAL-DESIGN.md` 先让内容取舍和视觉方向可审阅，再由正式 `ExplainerDocument` IR 通过 `ExplanationBeat` 绑定旁白短语与视觉动作。
 
@@ -14,15 +14,15 @@ Seqvio 为 coding agent 提供从真实系统捕获到讲解视频的完整路
 
 ## Demo
 
-预渲染产品介绍视频，使用 CosyVoice 旁白，覆盖三个视觉风格包（`@seqvio/whiteboard`、`@seqvio/scatterbrain`、`@seqvio/product-demo`）。源 composition 位于 [`examples/compositions/`](./examples/compositions/)。
+当前 720p 产品演示以 native-module CI 故障诊断为例，展示可审阅策划、
+正式 `ExplainerDocument`、短语级 `ExplanationBeat`、QA 和本地渲染路径。
 
-**英文介绍** — [`seqvio-overview-en.tsx`](./examples/compositions/seqvio-overview-en.tsx)
+**[观看当前带旁白演示](./docs/assets/videos/seqvio-product-hunt-en.mp4)**
+— 源码：[`seqvio-product-hunt-en.tsx`](./examples/compositions/seqvio-product-hunt-en.tsx)
 
-https://github.com/user-attachments/assets/83687d9c-63f0-4544-a67a-8f6eacc19928
-
-**中文介绍** — [`seqvio-overview-zh.tsx`](./examples/compositions/seqvio-overview-zh.tsx)
-
-https://github.com/user-attachments/assets/3ce605bc-7ad1-449b-a67c-5d8368f5398b
+中英文 overview composition 仍保留在
+[`examples/compositions/`](./examples/compositions/)；此前发布的视频早于
+当前解释契约，因此不再作为主要产品演示。
 
 ## 快速开始
 
@@ -60,7 +60,9 @@ npm install -g @seqvio/renderer
 seqvio-render --help
 ```
 
-已发布包：`@seqvio/core`、`@seqvio/whiteboard`、`@seqvio/scatterbrain`、`@seqvio/product-demo`、`@seqvio/technical`、`@seqvio/renderer`。
+Public 包：`@seqvio/core`、`@seqvio/whiteboard`、`@seqvio/scatterbrain`、
+`@seqvio/product-demo`、`@seqvio/technical`、`@seqvio/renderer`。实验性捕获包：
+`@seqvio/capture`、`@seqvio/browser-recorder`、`@seqvio/terminal-narrator`。
 
 当 composition 直接 import 可选视觉包时，可额外安装：
 
@@ -96,7 +98,10 @@ export ELEVENLABS_API_KEY=your_key
 
 > 使用 `/seqvio`，先生成并审阅讲解策划与视觉设计说明，再编译成 4 场景中文 ExplainerDocument，加入短语锚定视觉 Beat，运行 QA 后渲染 MP4。
 
-Skill 会引导 agent：选示例 composition、改 TSX、提取旁白元数据、合成音频、运行 `seqvio-render`。
+推荐路径会先审阅 `EDITORIAL.md` 和 `VISUAL-DESIGN.md`，编译
+`ExplainerDocument`，解析旁白时间，运行 `seqvio-qa`，最后渲染 MP4。
+需要精细视觉控制时仍可直接手写 TSX；实验性的 terminal/browser 适配器
+可以向同一个 IR 提供真实观察场景，但不是创作型讲解的必需依赖。
 
 支持 Cursor、Claude Code、Codex、Gemini CLI 等支持 skills 的 coding agent。
 
@@ -188,7 +193,7 @@ Skill 主文件：[`skills/seqvio/SKILL.md`](./skills/seqvio/SKILL.md)，参考
 | [`audio-workflow.md`](./skills/seqvio/references/audio-workflow.md) | 提取、合成、混流旁白 |
 | [`render-workflow.md`](./skills/seqvio/references/render-workflow.md) | build、render、smoke test 命令 |
 | [`production-techniques.md`](./skills/seqvio/references/production-techniques.md) | voice-first timing、参考风格分析和视觉 QA 规则 |
-| [`planning-workflow.md`](./skills/seqvio/references/planning-workflow.md) | storyboard IR planning 和 agent handoff |
+| [`planning-workflow.md`](./skills/seqvio/references/planning-workflow.md) | Editorial/Visual 规划和 agent handoff |
 
 安装 skill（见 [快速开始](#快速开始)）：
 
@@ -222,7 +227,7 @@ Seqvio 是 coding agent 用来解释想法的视觉语言，而不只是一个
 - Browser 场景支持录制视频、光标/聚焦/点击元数据、真实动作时钟和 time-mapped 媒体 seek
 - `ExplanationBeat` cue、精确短语锚点、视觉动作、捕获证据、TTS 后 `outputFrame` 和语义 `sceneTimings[].timeMap`
 - `@seqvio/core`：`VideoComposition`、`Scene`、`Transition`
-- ExplainerDocument 与保留的 Storyboard IR schema、validation、pacing 和 TSX compile helpers
+- ExplainerDocument 是正式 IR；Storyboard IR 仅作为白板输入兼容路径保留
 - `seqvio-render`：TSX 到 MP4
 - `seqvio-audio`：manifest 提取与 TTS 合成
 - `seqvio-qa`：baseline/capture profile、稳定音频/时序/媒体诊断、警告升级和关键帧视觉检查
@@ -295,6 +300,7 @@ node packages/renderer/dist/cli.js \
 | [`@seqvio/technical`](./packages/technical) | 技术讲解 runtime：代码走读、架构图、终端演示、标注和内置字体 |
 | [`@seqvio/terminal-narrator`](./packages/terminal-narrator) | Pre-stable node-pty/xterm 捕获 → IR/ExplanationBeat → 可选旁白 MP4 |
 | [`@seqvio/browser-recorder`](./packages/browser-recorder) | Pre-stable Chromium action 捕获，保留真实动作时间 → IR/ExplanationBeat |
+| [`@seqvio/capture`](./packages/capture) | 实验性的共享 capture session 和 artifact 契约 |
 | [`@seqvio/renderer`](./packages/renderer) | TSX bundler，以及 `seqvio-render` / `seqvio-audio` CLI |
 
 ## 文档
@@ -310,6 +316,8 @@ node packages/renderer/dist/cli.js \
 - [`examples/compositions/README.md`](./examples/compositions/README.md)
 - [`skills/seqvio/SKILL.md`](./skills/seqvio/SKILL.md)
 - [`skills/seqvio/references/production-techniques.md`](./skills/seqvio/references/production-techniques.md)
+- [`docs/marketing/POSITIONING.md`](./docs/marketing/POSITIONING.md) — 当前产品定位和能力边界
+- [`docs/marketing/FEATURE-STATUS.md`](./docs/marketing/FEATURE-STATUS.md) — public 与 experimental 的对外措辞
 
 如果文档与代码冲突，以代码和 [`docs/COMPOSITION-AUTHORING.md`](./docs/COMPOSITION-AUTHORING.md) 为准。
 
diff --git a/docs/README.md b/docs/README.md
index d77b134..5b6a87a 100644
--- a/docs/README.md
+++ b/docs/README.md
@@ -21,6 +21,8 @@ If you are trying to use Seqvio today, read these first:
 | [`../examples/compositions/README.md`](../examples/compositions/README.md) | Composition examples and render command |
 | [`../packages/whiteboard/README.md`](../packages/whiteboard/README.md) | Whiteboard component API and examples |
 | [`../packages/product-demo/package.json`](../packages/product-demo/package.json) | Product walkthrough component package metadata |
+| [`marketing/POSITIONING.md`](./marketing/POSITIONING.md) | Public positioning and product boundary |
+| [`marketing/FEATURE-STATUS.md`](./marketing/FEATURE-STATUS.md) | Public and experimental capability wording |
 
 ## Current Source of Truth
 
@@ -48,6 +50,9 @@ These docs describe the repository as it exists today:
 | [`VISION.md`](./VISION.md) | Product positioning, target users, scope, and design principles (slow-changing source of truth) |
 | [`ROADMAP.md`](./ROADMAP.md) | Directional bet and phase ordering; what to build next and why. Subordinate to `VISION.md` |
 | [`EXPLAINER-FRAMEWORK-IMPROVEMENT-PLAN.md`](./EXPLAINER-FRAMEWORK-IMPROVEMENT-PLAN.md) | Active implementation plan for capture, ExplanationBeat timing, deterministic playback, QA, and release readiness |
+| [`marketing/LANDING-COPY.md`](./marketing/LANDING-COPY.md) | Versioned landing page copy |
+| [`marketing/PRODUCT-HUNT.md`](./marketing/PRODUCT-HUNT.md) | Product Hunt tagline, description, and demo promise |
+| [`marketing/DEMO-SCRIPT.md`](./marketing/DEMO-SCRIPT.md) | Current 60-75 second product demo sequence |
 
 Active, time-boxed work items live in GitHub issues and milestones, not in a docs file.
 
diff --git a/docs/assets/demo/README.md b/docs/assets/demo/README.md
index 9e88824..e6ff9bb 100644
--- a/docs/assets/demo/README.md
+++ b/docs/assets/demo/README.md
@@ -4,6 +4,7 @@ README demo videos now live in [`../videos/`](../videos/):
 
 - `seqvio-overview-en.mp4` + `seqvio-overview-en.preview.gif`
 - `seqvio-overview-zh.mp4` + `seqvio-overview-zh.preview.gif`
+- `seqvio-product-hunt-en.mp4` (current explanation-contract demo)
 
 GitHub README cannot inline-play repo-hosted MP4 with `<video>`; README uses the GIF previews (click through to MP4 for narration).
 
@@ -18,6 +19,16 @@ Source compositions:
 
 - `examples/compositions/seqvio-overview-en.tsx`
 - `examples/compositions/seqvio-overview-zh.tsx`
+- `examples/compositions/seqvio-product-hunt-en.tsx`
+
+The overview and Product Hunt sources must follow the current product path in
+[`../../marketing/POSITIONING.md`](../../marketing/POSITIONING.md). Do not use
+the retained Storyboard compatibility input as the primary workflow.
+
+For tracked demo video, render at 1280x720 or 1920x1080 with the repository's
+medium-or-higher quality profile. Bitrate varies substantially for flat motion
+graphics, so inspect text-heavy key frames and the encoded output instead of
+using file size or average bitrate as the only acceptance criterion.
 
 Local CLI renders and audio intermediates should write to `output/` at the repository root (gitignored).
 
diff --git a/docs/assets/videos/seqvio-product-hunt-en.mp4 b/docs/assets/videos/seqvio-product-hunt-en.mp4
new file mode 100644
index 0000000..c2ca3f8
Binary files /dev/null and b/docs/assets/videos/seqvio-product-hunt-en.mp4 differ
diff --git a/docs/marketing/DEMO-SCRIPT.md b/docs/marketing/DEMO-SCRIPT.md
new file mode 100644
index 0000000..0341e8c
--- /dev/null
+++ b/docs/marketing/DEMO-SCRIPT.md
@@ -0,0 +1,42 @@
+# Current Product Demo Script
+
+Target length: 60-75 seconds. Use 1920x1080 for Product Hunt and 1280x720
+for the README variant. Keep code and terminal text readable at normal playback.
+
+## 1. Failure, 0-8s
+
+Show the observed native-module CI error. State the question: why did a rebuild
+report success while the module still failed to load?
+
+## 2. Editorial decision, 8-18s
+
+Show `EDITORIAL.md`. Select `causal-diagnosis` as the primary soft pattern and
+`evidence-demonstration` as supporting advice. Highlight the explicit omission
+of unrelated release details.
+
+## 3. Visual direction, 18-27s
+
+Show `VISUAL-DESIGN.md`: terminal for observed output, diagram for the load path,
+and code only for the configuration that caused install scripts to be blocked.
+
+## 4. Executable explanation, 27-40s
+
+Show `explainer.json`. Connect one narrated claim to an `ExplanationBeat` with a
+phrase anchor, visual target, and captured step id. Clearly distinguish authored
+diagram content from capture-derived terminal evidence.
+
+## 5. Timing and QA, 40-55s
+
+Resolve the spoken phrase after TTS. Show the resulting output frame and a short
+QA report covering timing, evidence, media, and key-frame checks.
+
+## 6. Result, 55-68s
+
+Play the finished explanation: symptom, expected load path, blocked install
+script, corrected configuration, and verified module load.
+
+## 7. Close, 68-75s
+
+End on: "Turn real technical work into explainable video." Show the GitHub URL
+and `npm install -g @seqvio/renderer`. Label terminal/browser capture as
+experimental in supporting copy, not in the spoken close.
diff --git a/docs/marketing/FEATURE-STATUS.md b/docs/marketing/FEATURE-STATUS.md
new file mode 100644
index 0000000..96e55d7
--- /dev/null
+++ b/docs/marketing/FEATURE-STATUS.md
@@ -0,0 +1,15 @@
+# Feature Status For Public Materials
+
+| Area | Status | Public wording |
+| --- | --- | --- |
+| `ExplainerDocument` and five scene compilers | Public | Complete authoring and rendering path |
+| EditorialPlan / VisualDesignBrief | Public | Reviewable planning artifacts before executable IR |
+| ExplanationBeat timing and post-TTS alignment | Public | Narration, visual action, and evidence share one timing contract |
+| `seqvio-qa` | Public | Deterministic audio, pacing, media, semantic, and key-frame checks |
+| `@seqvio/terminal-narrator` | Experimental | Real terminal capture with platform-specific prerequisites |
+| `@seqvio/browser-recorder` | Experimental | Chromium action capture with recorded evidence and timing |
+| `@seqvio/capture` | Experimental | Shared capture contracts and artifacts |
+| Screenshot privacy masking | Deferred | Do not promise this in marketing copy |
+
+Marketing demos must identify whether a scene is authored or capture-derived.
+Do not describe experimental adapters as a universal one-click workflow.
diff --git a/docs/marketing/LANDING-COPY.md b/docs/marketing/LANDING-COPY.md
new file mode 100644
index 0000000..76bbc38
--- /dev/null
+++ b/docs/marketing/LANDING-COPY.md
@@ -0,0 +1,43 @@
+# Landing Page Copy
+
+## Hero
+
+### Seqvio
+
+### Turn real technical work into explainable video.
+
+Seqvio gives coding agents a visual language for turning technical ideas and
+real terminal/browser activity into narration-locked explainer videos.
+
+**Primary CTA:** Watch the workflow demo
+**Secondary CTA:** Install the agent skill
+
+Open source. Local rendering. Reviewable plans. Deterministic QA.
+
+## How it works
+
+1. Plan the explanation in readable `EDITORIAL.md` and `VISUAL-DESIGN.md` files.
+2. Compile authored content or observed capture into `ExplainerDocument`.
+3. Bind spoken phrases, visual actions, and evidence with `ExplanationBeat`.
+4. Resolve TTS timing, run QA, and render a local MP4.
+
+## Why it is different
+
+Most generated videos animate an idea from a prompt. Seqvio can preserve what
+actually happened in a terminal or browser session, then explain that evidence
+with the same timing contract used for authored scenes.
+
+## Current scope
+
+Use the public authoring path for whiteboard, code, and diagram explainers.
+Terminal and browser capture are available as experimental adapters and require
+their host runtimes. See the repository's feature status before adopting them
+in CI.
+
+## What you get
+
+- Human-readable editorial and visual decisions.
+- A formal, inspectable `ExplainerDocument`.
+- Phrase-level narration and visual alignment.
+- Deterministic local rendering and machine-readable QA.
+- Reusable TSX compositions for technical content pipelines.
diff --git a/docs/marketing/POSITIONING.md b/docs/marketing/POSITIONING.md
new file mode 100644
index 0000000..3ebaff3
--- /dev/null
+++ b/docs/marketing/POSITIONING.md
@@ -0,0 +1,55 @@
+# Seqvio Positioning
+
+## One sentence
+
+Seqvio turns technical ideas and real terminal/browser evidence into narration-locked explainer videos for coding agents.
+
+## What it is
+
+Seqvio is an open-source, local explainer production framework. A host agent or
+human author makes the creative decisions; Seqvio provides the reviewable
+authoring artifacts, executable `ExplainerDocument` contract, visual scene
+vocabulary, timing alignment, capture adapters, QA, and MP4 rendering.
+
+## The product path
+
+```text
+source material or real capture
+  -> EditorialPlan
+  -> VisualDesignBrief
+  -> ExplainerDocument
+  -> ExplanationBeat timing
+  -> TTS alignment and QA
+  -> local MP4
+```
+
+`EditorialPlan` and `VisualDesignBrief` are human-readable planning surfaces.
+`ExplainerDocument` is the formal execution IR. `ExplanationBeat` keeps a
+spoken phrase, the visual action, and capture evidence in one unit.
+
+## Current capability boundary
+
+Public authoring and rendering support five complete scene families:
+`whiteboard`, `code`, `diagram`, `terminal`, and `browser`. Terminal and browser
+scenes are capture-derived and must not be invented as if they were observed.
+The capture adapters are experimental even though their compiled scene paths
+and QA contracts are implemented.
+
+The six explanation patterns are optional editorial advice: `causal-diagnosis`,
+`mechanism-trace`, `system-flow`, `evidence-demonstration`,
+`misconception-reframe`, and `progressive-model`. They are not templates and do
+not constrain the executable IR.
+
+## What Seqvio is not
+
+- A general-purpose nonlinear editor.
+- A photorealistic or one-shot prompt-to-video generator.
+- A hosted planner or LLM provider.
+- A guarantee of high aesthetic quality without human editorial and visual review.
+
+## Audience
+
+The primary audience is developers, developer advocates, technical writers,
+educators, and agents producing technical explainers inside a versioned code
+repository. Product walkthroughs and broader educational videos are supported,
+but the strongest differentiation is fidelity to real system activity.
diff --git a/docs/marketing/PRODUCT-HUNT.md b/docs/marketing/PRODUCT-HUNT.md
new file mode 100644
index 0000000..2431148
--- /dev/null
+++ b/docs/marketing/PRODUCT-HUNT.md
@@ -0,0 +1,28 @@
+# Product Hunt Copy
+
+## Tagline
+
+Open-source explainer videos for coding agents.
+
+## Short description
+
+Seqvio turns technical ideas and real terminal/browser evidence into
+narration-locked explainer videos. Review the editorial plan, choose a visual
+direction, compile an inspectable `ExplainerDocument`, align each spoken phrase
+to a visual action, run QA, and render locally.
+
+## Maker comment
+
+Seqvio is for developers and technical educators who want video production to
+live beside the code it explains. A host agent makes the creative decisions;
+Seqvio makes them reviewable and executable. The same `ExplanationBeat` can
+connect a narrated claim to a diagram, code highlight, or observed terminal and
+browser state. Capture adapters are experimental and clearly labeled; the public
+authoring and rendering path is deterministic and local.
+
+## Demo promise
+
+The first 75 seconds should show one real technical problem from start to finish:
+the readable plan, the observed evidence, the formal IR, one phrase-level beat,
+the QA result, and the final video. A style gallery can follow as supporting
+material; it should not be the product's primary proof.
diff --git a/examples/compositions/seqvio-overview-en.tsx b/examples/compositions/seqvio-overview-en.tsx
index cebc12a..d02a331 100644
--- a/examples/compositions/seqvio-overview-en.tsx
+++ b/examples/compositions/seqvio-overview-en.tsx
@@ -47,7 +47,7 @@ function IntroScene() {
         duration={14}
         style={{ padding: '14px 20px' }}
       >
-        <span style={{ fontSize: TS.caption }}>SEQVIO · V0.5</span>
+        <span style={{ fontSize: TS.caption }}>SEQVIO · V0.8</span>
       </StickyNote>
 
       <Scrawl
@@ -70,7 +70,7 @@ function IntroScene() {
       />
 
       <StickyNote
-        title="prompt.md"
+        title="source.md"
         position={{ x: 772, y: 88 }}
         width={300}
         color="yellow"
@@ -83,7 +83,7 @@ function IntroScene() {
         lessons, product walkthroughs, concepts
       </StickyNote>
       <StickyNote
-        title="storyboard.json"
+        title="EDITORIAL.md"
         position={{ x: 884, y: 300 }}
         width={312}
         color="green"
@@ -93,7 +93,7 @@ function IntroScene() {
         start={76}
         duration={16}
       >
-        scenes, narration, visual intent, timing
+        objective, evidence, explanation structure
       </StickyNote>
       <Polaroid
         caption="MP4 · rendered locally"
@@ -376,7 +376,7 @@ function WorkflowScene() {
     >
       <ProductTitle
         title="@seqvio/product-demo"
-        subtitle="Agent produces structure; renderer validates, aligns narration, renders locally."
+        subtitle="Plans become ExplainerDocument; Beats align voice, visuals, and evidence."
         position={{ x: 64, y: 56 }}
         start={0}
       />
@@ -389,7 +389,7 @@ function WorkflowScene() {
         title="Seqvio Render Console"
         start={36}
       >
-        <ScreenshotPlaceholder label="prompt.md -> storyboard.json" start={52} />
+        <ScreenshotPlaceholder label="EDITORIAL.md -> explainer.json" start={52} />
       </BrowserFrame>
 
       <CursorPath
@@ -403,7 +403,7 @@ function WorkflowScene() {
         duration={62}
       />
       <Callout
-        text="Storyboard IR is generated, validated, repaired, and compiled deterministically."
+        text="Editorial and visual plans are reviewed, compiled into ExplainerDocument, and repaired before deterministic rendering."
         position={{ x: 848, y: 262 }}
         width={318}
         start={118}
@@ -481,7 +481,7 @@ function ClosingScene() {
         color="#22C55E"
       />
       <Callout
-        text="seqvio-render fuses TSX, the CosyVoice track, and local rendering into one MP4."
+        text="seqvio-qa checks timing and key frames before seqvio-render produces the local MP4."
         position={{ x: 768, y: 300 }}
         width={392}
         start={96}
@@ -562,17 +562,17 @@ export const meta: RenderableMeta = {
       {
         id: 'whiteboard',
         sceneId: 'whiteboard',
-        text: 'Seqvio gets its visuals from several style packages. The whiteboard package gives you handwritten text, sketch-like shapes, and pen timing, ideal for explaining a concept or a workflow step by step.',
+        text: 'Seqvio starts with a reviewable editorial plan and visual direction. Whiteboard scenes provide handwritten text, sketch-like shapes, and pen timing when a concept needs to unfold step by step.',
       },
       {
         id: 'scatterbrain',
         sceneId: 'scatterbrain',
-        text: 'The scatterbrain package adds sticky-note and cork-board workshop scenes for brainstorming and shaping ideas. Different look, but the same timing, narration, and rendering contracts.',
+        text: 'The visual vocabulary also includes workshop, product, code, and diagram scenes. They share one timing and narration contract, so the agent can choose the form that best serves each part of the explanation.',
       },
       {
         id: 'workflow',
         sceneId: 'workflow',
-        text: 'The product-demo package covers product walkthroughs. The whole flow is deterministic: an agent produces the structured storyboard, and the renderer validates it, aligns narration, and renders locally, so the visuals follow the synthesized voice.',
+        text: 'The product-demo package covers product walkthroughs, while technical scenes explain code and diagrams. The flow is deterministic: plans become an ExplainerDocument, spoken phrases bind to visual actions, QA checks the result, and the renderer produces the video locally.',
       },
       {
         id: 'closing',
diff --git a/examples/compositions/seqvio-overview-shared.tsx b/examples/compositions/seqvio-overview-shared.tsx
index fa50a82..d80bd23 100644
--- a/examples/compositions/seqvio-overview-shared.tsx
+++ b/examples/compositions/seqvio-overview-shared.tsx
@@ -100,6 +100,24 @@ export interface OverviewProps {
   stageHeight?: number;
 }
 
+function ScaledOverviewStage({ children, scale }: { children: ReactNode; scale: number }) {
+  return (
+    <div
+      style={{
+        position: 'absolute',
+        left: 0,
+        top: 0,
+        width: W,
+        height: H,
+        transform: `scale(${scale})`,
+        transformOrigin: '0 0',
+      }}
+    >
+      {children}
+    </div>
+  );
+}
+
 function clamp(value: number): number {
   return Math.max(0, Math.min(1, value));
 }
@@ -414,8 +432,8 @@ function RagExplanationScene({ copy, duration, enhanced }: { copy: OverviewCopy;
   const boardY = (H - boardH) / 2;
   const clipNow = Math.min(9, Math.floor(Math.min(1, frame / drawDuration) * 9));
   const actionLabels = copy.lang === 'zh'
-    ? ['提问', '向量', '检索', '回答']
-    : ['ASK', 'VECTOR', 'RETRIEVE', 'RESPOND'];
+    ? ['观察', '追踪', '诊断', '验证']
+    : ['OBSERVE', 'TRACE', 'DIAGNOSE', 'VERIFY'];
   const icons = ['lightbulb', 'plus', 'document', 'check'];
   return (
     <div
@@ -639,7 +657,7 @@ function MiniWhiteboard({ copy }: { copy: OverviewCopy }) {
       singlePen={false}
     >
       <DrawText
-        text="RAG"
+        text="MODEL"
         position={{ x: 34, y: 48 }}
         fontSize={38}
         fontWeight="bold"
@@ -711,7 +729,7 @@ function MiniScatter({ copy }: { copy: OverviewCopy }) {
         duration={20}
         style={{ padding: 12 }}
       >
-        RAG
+        PLAN
       </StickyNote>
       <Doodle
         type="arrow"
@@ -733,19 +751,19 @@ function MiniProductDemo() {
         position={{ x: 18, y: 38 }}
         width={308}
         height={248}
-        url="rag-demo.local"
-        title="sample output"
+        url="ci-run.local"
+        title="captured evidence"
         start={18}
         duration={22}
       >
         <div style={{ padding: 18, fontFamily: UI_STACK }}>
           <div style={{ fontSize: 17, fontWeight: 800, color: C.ink }}>
-            How does RAG answer?
+            Native module load
           </div>
           <div style={{ marginTop: 16, height: 10, width: 230, background: '#DCE5F3' }} />
           <div style={{ marginTop: 9, height: 10, width: 186, background: '#DCE5F3' }} />
           <div style={{ marginTop: 24, display: 'flex', gap: 8 }}>
-            {['Q', 'DB', 'A'].map((item, index) => (
+            {['BUILD', 'PTY', 'OK'].map((item, index) => (
               <div
                 key={item}
                 style={{
@@ -765,7 +783,7 @@ function MiniProductDemo() {
         </div>
       </BrowserFrame>
       <Callout
-        text="context found"
+        text="load verified"
         position={{ x: 182, y: 236 }}
         width={132}
         start={92}
@@ -785,13 +803,13 @@ function StylesScene({ copy, duration, enhanced }: { copy: OverviewCopy; duratio
   ];
   return (
     <LightStage dense={enhanced}>
-      {enhanced ? <SceneMeta index="05" label="VISUAL LANGUAGES" light /> : null}
+      {enhanced ? <SceneMeta index="05" label="EXPLANATION CONTRACT" light /> : null}
       <div
         style={{
           position: 'absolute',
           left: 74,
           top: 102,
-          fontSize: 50,
+          fontSize: enhanced ? 42 : 50,
           fontWeight: 850,
           ...reveal(frame, 4, 22),
         }}
@@ -859,7 +877,7 @@ function StylesScene({ copy, duration, enhanced }: { copy: OverviewCopy; duratio
 function ProofScene({ copy, duration, enhanced }: { copy: OverviewCopy; duration: number; enhanced: boolean }) {
   const frame = useCurrentFrame() * 246 / Math.max(1, duration);
   const commands = [
-    'seqvio-generate validate storyboard.json',
+    'seqvio-generate validate --ir explainer.json --ir-format explainer',
     'seqvio-audio synthesize --manifest audio.json',
     'seqvio-qa --frames 0,120,240',
     'seqvio-render --component rag-explainer.tsx',
@@ -1157,25 +1175,27 @@ export function SeqvioOverview({
       backgroundColor={C.navy}
       audio={audio}
     >
-      <div
-        style={{
-          position: 'absolute',
-          left: 0,
-          top: 0,
-          width: W,
-          height: H,
-          transform: `scale(${scale})`,
-          transformOrigin: '0 0',
-        }}
-      >
-        <Scene id="hook" duration={hook}><HookScene copy={copy} duration={hook} enhanced={enhanced} /></Scene>
-        <Scene id="promise" duration={promise}><PromiseScene copy={copy} duration={promise} enhanced={enhanced} /></Scene>
-        <Scene id="prompt" duration={prompt}><PromptScene copy={copy} duration={prompt} enhanced={enhanced} /></Scene>
-        <Scene id="explanation" duration={explanation}><RagExplanationScene copy={copy} duration={explanation} enhanced={enhanced} /></Scene>
-        <Scene id="styles" duration={styles}><StylesScene copy={copy} duration={styles} enhanced={enhanced} /></Scene>
-        <Scene id="proof" duration={proof}><ProofScene copy={copy} duration={proof} enhanced={enhanced} /></Scene>
-        <Scene id="closing" duration={closing}><ClosingScene copy={copy} duration={closing} enhanced={enhanced} /></Scene>
-      </div>
+      <Scene id="hook" duration={hook}>
+        <ScaledOverviewStage scale={scale}><HookScene copy={copy} duration={hook} enhanced={enhanced} /></ScaledOverviewStage>
+      </Scene>
+      <Scene id="promise" duration={promise}>
+        <ScaledOverviewStage scale={scale}><PromiseScene copy={copy} duration={promise} enhanced={enhanced} /></ScaledOverviewStage>
+      </Scene>
+      <Scene id="prompt" duration={prompt}>
+        <ScaledOverviewStage scale={scale}><PromptScene copy={copy} duration={prompt} enhanced={enhanced} /></ScaledOverviewStage>
+      </Scene>
+      <Scene id="explanation" duration={explanation}>
+        <ScaledOverviewStage scale={scale}><RagExplanationScene copy={copy} duration={explanation} enhanced={enhanced} /></ScaledOverviewStage>
+      </Scene>
+      <Scene id="styles" duration={styles}>
+        <ScaledOverviewStage scale={scale}><StylesScene copy={copy} duration={styles} enhanced={enhanced} /></ScaledOverviewStage>
+      </Scene>
+      <Scene id="proof" duration={proof}>
+        <ScaledOverviewStage scale={scale}><ProofScene copy={copy} duration={proof} enhanced={enhanced} /></ScaledOverviewStage>
+      </Scene>
+      <Scene id="closing" duration={closing}>
+        <ScaledOverviewStage scale={scale}><ClosingScene copy={copy} duration={closing} enhanced={enhanced} /></ScaledOverviewStage>
+      </Scene>
     </VideoComposition>
   );
 }
diff --git a/examples/compositions/seqvio-overview-zh.tsx b/examples/compositions/seqvio-overview-zh.tsx
index 1f273c1..c0a49d5 100644
--- a/examples/compositions/seqvio-overview-zh.tsx
+++ b/examples/compositions/seqvio-overview-zh.tsx
@@ -49,7 +49,7 @@ function IntroScene() {
         duration={14}
         style={{ padding: '14px 20px' }}
       >
-        <span style={{ fontSize: TS.caption }}>SEQVIO · V0.5</span>
+        <span style={{ fontSize: TS.caption }}>SEQVIO · V0.8</span>
       </StickyNote>
 
       <Scrawl
@@ -72,7 +72,7 @@ function IntroScene() {
       />
 
       <StickyNote
-        title="prompt.md"
+        title="source.md"
         position={{ x: 772, y: 88 }}
         width={300}
         color="yellow"
@@ -85,7 +85,7 @@ function IntroScene() {
         课程、产品介绍、技术概念
       </StickyNote>
       <StickyNote
-        title="storyboard.json"
+        title="EDITORIAL.md"
         position={{ x: 884, y: 300 }}
         width={312}
         color="green"
@@ -95,7 +95,7 @@ function IntroScene() {
         start={76}
         duration={16}
       >
-        场景、旁白、视觉意图、时间关系
+        目标、证据、讲解结构和取舍
       </StickyNote>
       <Polaroid
         caption="MP4 · rendered locally"
@@ -378,7 +378,7 @@ function WorkflowScene() {
     >
       <ProductTitle
         title="@seqvio/product-demo + 确定性工作流"
-        subtitle="agent 产出结构，renderer 负责校验、对齐旁白、本地渲染。"
+        subtitle="计划编译成 ExplainerDocument；Beat 对齐旁白、画面和证据。"
         position={{ x: 64, y: 56 }}
         start={0}
       />
@@ -392,7 +392,7 @@ function WorkflowScene() {
         chromeFontFamily={LONG_CANG_STACK}
         start={36}
       >
-        <ScreenshotPlaceholder label="prompt.md -> storyboard.json" start={52} />
+        <ScreenshotPlaceholder label="EDITORIAL.md -> explainer.json" start={52} />
       </BrowserFrame>
 
       <CursorPath
@@ -406,7 +406,7 @@ function WorkflowScene() {
         duration={62}
       />
       <Callout
-        text="Storyboard IR 能被 agent 生成、校验、修复，再确定性地编译。"
+        text="Editorial 和 Visual 计划先审阅，再编译成 ExplainerDocument，修复后确定性渲染。"
         position={{ x: 848, y: 262 }}
         width={318}
         start={118}
@@ -485,7 +485,7 @@ function ClosingScene() {
         color="#22C55E"
       />
       <Callout
-        text="seqvio-render 把 TSX、CosyVoice 音轨和本地渲染合成一个 MP4。"
+        text="seqvio-qa 先检查时间和关键帧，再由 seqvio-render 生成本地 MP4。"
         position={{ x: 768, y: 300 }}
         width={392}
         start={96}
@@ -566,17 +566,17 @@ export const meta: RenderableMeta = {
       {
         id: 'whiteboard',
         sceneId: 'whiteboard',
-        text: 'Seqvio 的视觉能力由多个风格包提供。whiteboard 包给你手写文字、草图形状和画笔节奏，适合一步步讲清楚一个概念或流程。',
+        text: 'Seqvio 先形成可审阅的讲解策划和视觉方向。需要逐步展开概念时，whiteboard 场景提供手写文字、草图形状和画笔节奏。',
       },
       {
         id: 'scatterbrain',
         sceneId: 'scatterbrain',
-        text: 'scatterbrain 包提供便签和软木板式的工作台场景，适合发散和梳理想法。风格不同，但时序、旁白和渲染契约完全一致。',
+        text: '视觉词汇还包括工作台、产品、代码和架构场景。它们共享同一套时间和旁白契约，让 agent 为讲解的每一部分选择最合适的形式。',
       },
       {
         id: 'workflow',
         sceneId: 'workflow',
-        text: 'product-demo 包用来做产品演示。整个流程是确定性的：agent 产出结构化的 storyboard，renderer 负责校验、对齐旁白、本地渲染，让画面时长跟随合成出来的语音。',
+        text: 'product-demo 包用来做产品演示，technical 场景则解释代码和架构。整个流程是确定性的：计划编译成 ExplainerDocument，旁白短语绑定视觉动作，QA 检查结果，再由 renderer 本地生成视频。',
       },
       {
         id: 'closing',
diff --git a/examples/compositions/seqvio-product-hunt-en.tsx b/examples/compositions/seqvio-product-hunt-en.tsx
index 7eb0b33..e1d7de6 100644
--- a/examples/compositions/seqvio-product-hunt-en.tsx
+++ b/examples/compositions/seqvio-product-hunt-en.tsx
@@ -15,37 +15,37 @@ const copy: OverviewCopy = {
   hookTop: 'A video can move',
   hookBottom: 'and still explain nothing.',
   hookRail: 'Motion is not the same as explanation.',
-  promiseTitle: 'A visual language for AI agents.',
-  promiseRail: 'Seqvio gives agents scenes, narration, and timed visual steps.',
-  vocabulary: ['SCENES', 'VOICE', 'VISUAL STEPS'],
-  promptLabel: 'A REAL TASK, INSIDE A CODING AGENT',
-  promptText: 'Using /seqvio, explain how RAG turns a question into an answer.',
-  promptRail: 'The host agent plans. Seqvio supplies the explainer vocabulary.',
-  files: ['storyboard.json', 'rag-explainer.tsx', 'narration manifest'],
-  ragTitle: 'How RAG turns a question into an answer',
-  ragSteps: ['QUESTION', 'EMBEDDING', 'CONTEXT', 'ANSWER'],
-  ragRail: 'The idea becomes a sequence people can follow.',
-  stylesTitle: 'One explanation. Three visual languages.',
-  stylesRail: 'Choose the visual grammar that fits the idea.',
-  styleLabels: ['WHITEBOARD LESSON', 'STICKY-NOTE WORKSHOP', 'PRODUCT WALKTHROUGH'],
-  styleNotes: ['step by step', 'question', 'context', 'answer'],
+  promiseTitle: 'Explain what really happened.',
+  promiseRail: 'Seqvio turns technical ideas and system evidence into narrated video.',
+  vocabulary: ['PLAN', 'EVIDENCE', 'BEATS'],
+  promptLabel: 'A REAL TECHNICAL TASK, INSIDE A CODING AGENT',
+  promptText: 'Using /seqvio, explain why a native module failed in CI and how the repair was verified.',
+  promptRail: 'The host agent makes the editorial decisions. Seqvio makes them executable and checkable.',
+  files: ['EDITORIAL.md', 'explainer.json', 'qa-report.json'],
+  ragTitle: 'Why the native module failed in CI',
+  ragSteps: ['SYMPTOM', 'LOAD PATH', 'ROOT CAUSE', 'VERIFIED'],
+  ragRail: 'The claim stays connected to the observed evidence.',
+  stylesTitle: 'The right evidence for each claim.',
+  stylesRail: 'Use authored scenes for models and capture-derived scenes for what actually happened.',
+  styleLabels: ['AUTHORED MODEL', 'REVIEWABLE PLAN', 'CAPTURE EVIDENCE'],
+  styleNotes: ['expected path', 'claim', 'evidence', 'repair'],
   proofTitle: 'Real workflow. Real output.',
   proofRail: 'Author, validate, align, check, and render with the tools that ship today.',
   checks: ['structure', 'voice timing', 'key frame', 'final render'],
   closeKicker: 'OPEN SOURCE / AGENT-NATIVE EXPLAINERS',
   closeTitle: 'Teach your agent to explain.',
-  closeRail: 'Not just generated motion. A clear visual explanation.',
+  closeRail: 'Not just generated motion. A reviewable explanation grounded in evidence.',
   cta: 'github.com/makesynt/seqvio',
   hookTeaser: 'MADE BY AN AGENT',
   outputBadge: 'RENDERED BY SEQVIO · UNEDITED',
-  playerFile: 'rag-explainer.mp4',
+  playerFile: 'native-module-ci.mp4',
   ctaInstall: 'npm install -g @seqvio/renderer',
   ctaStar: 'Star on GitHub',
   proofOutputs: [
-    'valid · 7 scenes · 4 timed visual steps',
-    '7 cues synthesized · 42.8s aligned',
-    '3 frames checked · 0 overlaps',
-    '1285 frames → output/seqvio-rag.mp4',
+    'plan reviewed · omissions preserved',
+    'beats resolved · voice is the clock',
+    'QA passed · evidence and highlights checked',
+    'local MP4 · reproducible render artifacts',
   ],
 };
 
@@ -68,22 +68,22 @@ const narration = [
   {
     id: 'explanation',
     sceneId: 'explanation',
-    text: 'The agent maps the idea into scenes, narration, and timed visual steps. The question becomes an embedding. The embedding finds context. The context shapes the answer.',
+    text: 'The agent reviews the explanation plan and visual direction. Seqvio compiles them into an ExplainerDocument, then binds each spoken phrase to a visual action and observed evidence.',
   },
   {
     id: 'styles',
     sceneId: 'styles',
-    text: 'One explanation. Three visual languages: a whiteboard lesson, a sticky-note workshop, or a product walkthrough.',
+    text: 'Use whiteboard, code, and diagram scenes for authored models. Use terminal or browser capture when the explanation depends on what actually happened.',
   },
   {
     id: 'proof',
     sceneId: 'proof',
-    text: 'The agent authors. Seqvio validates the structure, aligns the voice, checks key frames, and renders locally.',
+    text: 'Seqvio resolves post-TTS timing, checks semantic beats and key frames, reports capture problems, and renders the final video locally.',
   },
   {
     id: 'closing',
     sceneId: 'closing',
-    text: 'Not just generated motion. A clear explanation. Teach your agent to explain with Seqvio. Open source on GitHub.',
+    text: 'Not just generated motion. A reviewable explanation grounded in evidence. Teach your agent to explain with Seqvio. Open source on GitHub.',
   },
 ];