Menu
Skills How it works Pricing
Free Video skillKnowledge videos

Evidence Story Video

Build a question-led explanation from a source ledger, evidence and counterevidence, then render an editable video.

Original evidence diagrams with local synthetic narration · complete listening review pending 16:9 · 0:28

Download the editable example

What you get

  • A source ledger, script and editable video project
  • Rendered MP4 with original diagrams or supplied evidence

You will need

  • Node.js, FFmpeg and project-local render dependencies
  • Sourced material and rights-cleared assets
  • User-owned TTS services when selected; local supplied audio is a separate path

Good fit

  • History, technology and humanities questions with inspectable sources
  • Separating observed facts, interpretation and reconstruction

Not for

  • Invented archive footage, quotes or causal evidence
  • Automatic factual approval from a render success

You say

“Read the installed SKILL.md. Use my actual inputs, preserve evidence and source media, and produce the declared deliverables. Report missing inputs and unperformed service calls.”

After installing, say something like this to your agent with your own topic and assets. It asks only for what is missing.

FAQ

Do I need an account?

No. Copy the install prompt into your agent, or download the free ZIP.

What does it cost to run?

The Skill is free. It needs the tools listed under “You will need”. Any voice, model or media service you choose bills you directly; FeedSkills does not include provider credits.

Can I use the videos commercially?

Videos you make are yours to use, including commercially. The Skill package is licensed as: MIT. Fonts, music, footage and services you add keep their own terms.

Which agents does it work with?

Agents that read SKILL.md files, such as Codex or Claude Code. What FeedSkills has actually tested is listed under Tested scope below.

Tested scope

Version
0.1.2
Acceptance
Acceptance for this version is still pending. It remains installable; check what has been tested before relying on it.
Tested
Two source-ledger topics and one scene revision were rendered with local supplied narration. Per-scene audio timing, full decode and source snapshots passed. Provider calls, word-timed captions and complete independent listening acceptance remain unverified.
Runs on
Codex · macOS · see dependencies
License
MIT

Details

Build an explanation that can be checked

Start with a question, reliable sources and the evidence that could change the answer. Keep observed facts, interpretation and reconstruction separate. The workflow turns that research into a source ledger, a narrated argument and an editable video project.

Choose pictures and diagrams for what they explain: a source excerpt establishes evidence, a comparison exposes a difference, and a process diagram shows a relationship. Generated imagery must be labelled as illustration; it cannot stand in for an archive, quotation or historical proof.

Keep the source and the cut connected

Retain source references beside the relevant claims and scenes. When a claim changes, update its diagram, narration and captions together, then render a separate revision. Review counterevidence and uncertainty before making the conclusion stronger.

Local rendering needs Node.js, FFmpeg and project-local Remotion dependencies. Supply rights-cleared evidence assets and narration, or use your own selected voice service. Online TTS sends narration to that provider and may incur charges. Captions require real timing information; the workflow must report missing timing rather than invent it.

Review the actual delivery

Check the finished MP4, source project, narration and source ledger together. Technical evals detect malformed inputs, missing files and output failures. They do not establish that a source is reliable, an inference is justified, or the full film is clear and accurate.