/VideoEssay

metadata:
  short-description: "Create DA-narrated video essays from content and a stance, with argument, visuals, narration, rendering, and review. Use Video for recorded footage."
name: VideoEssay
version: 1.3.0
evals:
  deterministic:
    - command: "bun skills/VideoEssay/Tools/DesignAudit.ts --check style-doc"
      falsifier: "a number or colour exported from Tools/Theme.ts is missing from References/Style.md, so the design system and its documentation have drifted"
    - command: "bun skills/VideoEssay/Tools/DesignAudit.ts --run smoke --if-present --check fonts,layout,legibility,sync"
      falsifier: "on a machine that keeps a run named `smoke` in its runs root, a composed frame carries a fallback font, an off-centre or undersized insert, text outside title-safe or behind the badge, colliding labels, an empty frame, sub-floor type, or a build that runs ahead of its word; with no such run it prints `skipped:` and exits 0"
  manual:
    reason: "argument quality and rendered essay coherence require a reviewer to assess the completed artifact"
    falsifier: "the reviewed essay misstates its source or stance, has unsupported claims, or the rendered video lacks its approved narration or visual beats"
description: Turns content plus a stance into a video essay narrated by the DA, any post, idea or feature plus a length into a short DA-narrated explainer video, or the author's own audio reading of their blog post (a recording or their voice clone) into a finished video of that post, with scripted visuals rendered on HyperFrames, presence layer, rendered and read back. USE WHEN video essay, explainer video, short video about this post, N-minute video of my blog post, make a video about this, respond to this video, argue against this in a video, debate this, video response, DA video, narrated video, turn this essay into a video, react to this interview, make a video from my audio, I recorded myself reading the post, turn this post into a video, video from my voice clone, audio to video. NOT FOR the user's camera footage or clips (Video), audio-only podcast cuts (Video), one-off animation (HyperFrames).

VideoEssay

**Contents:** Ideal State · Intent Routing · Tools · Data & Config · Quality loop · Gotchas · Examples

The DA's own video format, in two modes, each a fully automated production that proves itself before the user sees it. Essay mode argues first, then renders. Voice mode (`run.json` `mode: "voice"`) turns the author's post, read by their recording, their voice clone or the DA reading their exact words (`run.json` `narrator` `recorded` | `clone` | `da`), into a video of the real post page scrolling with the spoken sentence highlighted, inserts directed from what the post argues (`LIFEOS/TOOLS/VideoMeaning`: MeaningMap, Direction, a two-vendor plan review), and the reader's portrait in the corner badge (the author's, or the DA's on a DA-read run).

Ideal State

• An essay run argues before it renders: a linted, steelmanned `Essay.md` with verbatim, correctly attributed quotes reaches the user at the checkpoint (unless the ask skips it), and only then becomes `Script.json` and a render.
• A voice run narrates only the author's own words, its page reads like the live post, and every page highlight and insert is timed to the paragraph being spoken.
• Every render reads itself back before anyone sees it: `Out/verify.json` all green, and the frames and the `Out/review/` studio pass read.
• The mp4 and thumbnail reach `~/Downloads/<slug>/` only through `Deliver.ts`, after a passing ArtDirection critique and Proof receipt on those exact bytes.
• Every run stops in review state; nothing uploads from this skill. Publishing is the user's separate call through `Video`.

Intent Routing

| Intent | Trigger | File | |----------|---------|------| | **Argue** | "write the argument", "essay first", "what's our take on this", the first step of every new run | `Intents/Argue.md` | | **Script** | "script it", "turn the essay into beats", "plan the visuals" | `Intents/Script.md` | | **Produce** | "produce it", "render the essay", "make the video from the script", re-renders after an edit | `Intents/Produce.md` | | **Full** | "respond to this video", "video essay on this", "argue X in a video", raw content plus a stance to argue | `Intents/Full.md` | | **Voice** | "make a video from my audio", "I recorded the post", "use my clone to read the post", a post plus a recording | `Intents/Voice.md` | | **Explainer** | "make an explainer video", "a short video about this post / idea / feature", "a 2-minute video of my latest post", any source plus a length with no stance to argue | `Intents/Explainer.md` |

`Full` is `Argue` → review checkpoint → `Script` → `Produce`. The checkpoint after the essay is the default: text is cheap to change and a render is not. The user skips it by saying so in the ask ("straight through", "no review, just make it"). `Voice` has no essay and no DA-written narration; its narration rule, stages and inserts are in `Intents/Voice.md`. `Explainer` writes an `Essay.md` with `kind: explainer` (narration only, no opposing quotes) at the length asked for, then runs `Script` and `Produce` straight through; its optional YouTube leg (thumbnail on the post's header image, full description, private upload) runs through `Video`, never this skill.

| File | Read when | |------|-----------| | `References/ScriptSchema.md` | writing or linting `Script.json` | | `References/Style.md` | composing frames, the thumbnail, or running DesignAudit (the house design system) | | `References/Tools.md` | running or debugging one stage tool directly, choosing a flag, or finding a run's files | | `Setup.md` | first run on a machine, any tool stops naming Setup.md, or a Config.json field's meaning or default |

Tools

`Tools/Run.ts` drives every run; the other stage tools, the run layout and the composition files are in `References/Tools.md`.

| Tool | Contract | |------|----------| | `Tools