boring-video

Boring video

Orchestrate content development and a progressively refined HyperFrames project as one chain. Content is established before it is adapted for speech or visuals; scene direction establishes rendered visual evidence, storyboard plans buildable frames, and HyperFrames owns production and playable review.

Intake

Resolve existing decisions from the request and project files. First classify the input:

• Content-driven: research, evaluation, analysis, opinion, tutorial, news, profile, or another narrated work whose value depends on a complete argument and may also become a blog article.
• Direct-production: a locked script, product promo, UI demonstration, music-led piece, short motion graphic, or another work that does not need an article to establish its meaning.

For content-driven work, locate the first missing approved content artifact. A supplied article counts only when the user explicitly confirms it as the content master. Store or adapt it as content/ARTICLE.md with this provenance:

---
source: external
origin: <URL or path>
origin_revision: <Git revision, source digest, or captured-at timestamp>
status: approved
approved_at: YYYY-MM-DD
approved_digest: sha256:<digest>
voice:
  language: ...
  skills: []
  documents: []
  unavailable: []
  status: resolved | assumed
---

Compute approved_digest as SHA-256 of the complete file after removing only the approved_digest: line. Do not manufacture an article merely to satisfy the chain for direct-production work; load /hyperframes and let its current intent layer own intake and workflow selection.

When this intake creates or adapts content/ARTICLE.md, read references/markdown-review.md and present that authoritative Markdown before starting /to-spec-beats.

For content-driven work, ask one compact question for any missing production choices:

1. Presentation: faceless, footage, screen capture, talking head, or mixed media?
2. Collaboration: storyboard reviews or agent-executed production with final preview?
3. Voice: HyperFrames voice, ListenHub TTS, user recording, or no narration?
4. Design: existing design truth, named direction, or agent-proposed direction?

Record the answers in the HyperFrames BRIEF.md. In agent-executed work, propose and record a design direction when the user leaves it open. Direct-production work skips this wrapper intake because /hyperframes owns that brief.

Chain

Before /to-scenes starts the renderable project, read references/hyperframes-review-project.md. Pass its project identity and review state forward through SCENES.md, STORYBOARD.md, and the HyperFrames project manifest; downstream skills consume those artifacts rather than this internal reference.

For content-driven work, start at the first missing or explicitly revised artifact:

1. /develop-topic → content/RESEARCH.md, experiment reports, and approved content/TOPIC.md
2. /to-article → approved content/ARTICLE.md
3. /to-spec-beats → BEATS.md
4. /to-narration → NARRATION.md
5. /to-scenes → SCENES.md, shared HyperFrames project, and rendered visual evidence referenced from the Markdown
6. /to-storyboard → HyperFrames-native STORYBOARD.md
7. /to-video → production continuation and Studio review in that project

Finish each skill's completion criterion before advancing. A downstream finding returns to the artifact that owns the decision, then propagates forward.

When the user begins with an approved complete article, recompute its digest before starting at /to-spec-beats. When they begin with later content-driven artifacts, recompute every present artifact digest, require each child to carry the current digest of its immediate parent, and verify that the chain traces to an approved content/ARTICLE.md; then resume at the first missing stage. Artifact existence alone is not resumable state. Direct-production inputs bypass this content chain and begin at /hyperframes; never reinterpret a locked script as permission to invent an article, thesis, or claims.

Handoff

Presentation and workflow are separate decisions. Collect the presentation intent without selecting a workflow from memory. /to-video reads the current /hyperframes contract and performs the route.

Done

Content-driven planning is complete when the applicable approved content artifacts, four planning artifacts, and referenced visual evidence agree and satisfy their owners. The wrapper run is complete when /to-video continues the reviewed HyperFrames project with the confirmed voice branch through the selected production workflow and HyperFrames Studio owns the playable timeline review.