agentleFS
aicontentdropaicontentdrop/claude-plugin/skills/video-agent-brief/SKILL.md

video-agent-brief

Skill0 starsChanged 18 days agoMITView on GitHub

Prepare a brief for Drop, the AI Content Drop video agent, as text, before anyone opens the website. Covers the one-sentence goal and the facts the agent will ask for (product or subject, audience, purpose, length, aspect, platform, references and assets the person owns), what the agent does on the website (research, concepts, a plan, a price for every render that waits for the person's yes, a score, a verdict and the Director's notes on each finished piece), what the person decides and what they never need to do, and where the brief is pasted. Use when someone wants the video agent to handle a whole piece of work from one brief, a campaign or several videos from one idea, something in the style of a clip or a still they own, or wants to know what to prepare before they go to the website. Text advice only; it starts nothing, makes no video and spends nothing.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use it when the person wants the whole job handled from one brief: a campaign, a week of videos, several videos from one idea, an ad built …

    line 26
  2. Distinctive:

    Every render is priced on the page before it runs: the step in plain words and the number of credits, charged only if the step succeeds.

    line 72
  3. Distinctive:

    If the person asks what a plan includes, send them to <https://aicontentdrop.com/plans> and say nothing more about plans here.

    line 105

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Preparing a brief for Drop, the video agent

Drop is the AI Content Drop video agent. A person gives it one goal on the website and it carries that goal through to finished pieces: it researches ads that already work in the category, proposes concepts, writes a plan, prices every render and waits for the person's yes before anything is made, then judges what came back. Every step and every piece lands on a board the person can open.

Nothing in this skill starts, follows or approves any of that. You prepare the brief in words, say what will happen on the website, and hand the brief to the person. You have no tool for the agent and need none. The person needs no key and no credential in Claude, pays for nothing inside Claude, and signs in nowhere but the website.

1. When this is the right skill

Use it when the person wants the whole job handled from one brief: a campaign, a week of videos, several videos from one idea, an ad built on what competitors run, something in the style of a clip or a still they own, or when they ask what to prepare before they go to the website.

A single clip from one prompt is a smaller job: write the prompt with video-prompt-writer and choose the model with ai-video-model-picker. A short vertical drama is short-drama-writer.

2. Write the brief

The brief has two parts: the goal in one sentence, and the facts the agent will ask for before it plans. A goal the agent can act on names the subject, the length, the format and the purpose: "a 15-second vertical launch ad for a ceramic pour-over kettle, three shots, no voiceover" rather than "a kettle ad".

Then the facts. Fill in what the person told you, say which are missing, and ask only for the missing ones that change the plan:

Goal: one sentence
Product or subject: what the video is about, named precisely
Audience: who it is for
Purpose: launch, retarget, teach, entertain
Length: seconds
Aspect: vertical, square or wide
Platform: where it will run
References: a clip or a still the person owns, whose look the piece follows
Assets: a product image, a presenter image, a logo
Avoid: anything the video must not show or say

A finished product ad needs a product image and a presenter image; say so, and say that the agent asks for whatever it lacks. Keep the brief to what the person knows: do not invent a product fact, a claim or an audience, and do not choose a model. The agent chooses and prices each render on the website.

3. What happens on the website

Say this in plain words, so the person knows what they are walking into:

  • The run moves through phases: intake, research, concept, plan, produce, done. Research, concepts, the plan and the board cost nothing.
  • The agent may ask questions before it plans. The person answers them on the page; nobody answers for them.
  • Every render is priced on the page before it runs: the step in plain words and the number of credits, charged only if the step succeeds. Nothing is made until the person says yes to that step.
  • A clip or a still the person owns can be read into words first. Before a video is read, the page asks how the person holds the rights and offers three answers: it is my own work, I hold a licence, I have permission. State that question as the website's question; never answer it for the person, never suggest an answer, and never say a reference can be read without it. Reading costs no credits and is limited per day. The words it produces (the kind of reference, the shots, the cast, the look, any lines) go on the brief as "What I saw", and the person may correct them. The result recreates the look with an original cast; it never uses the reference's people, its lines or its brand.
  • Each finished piece comes back with a score from 0 to 100, a verdict (accept, retry, switch, accepted with notes, or unscored) with its reasons, and the Director's notes, which name the direction rules the agent applied and why. A retry or a switch is an offer of another take, priced on its own and made only after the person's yes; a declined offer leaves the piece as it is. The score is the agent's opinion in words, never a reason to approve anything.
  • Credits are charged only when a step succeeds. A failed or refused step costs nothing.

Quote no price. This skill holds no price for any step; the number is on the page, for the exact step, before it runs.

4. What the person decides, and what they never need to do

The person decides the goal, the answers to the agent's questions, every yes to a priced step, and whether to take an offered retry. They never need a key or a credential in Claude, never pay for anything inside Claude, and never have to sign in anywhere but the website.

If the person asks what a plan includes, send them to https://aicontentdrop.com/plans and say nothing more about plans here.

5. Hand it over

End with the brief, ready to paste, and one link: https://aicontentdrop.com/make. Say what opening it does: the page takes nothing from the link, so the brief is pasted on the page; a signed-out visitor is asked to sign in first; every render is priced there and waits for the person's yes. Sign-in and every account question happen on that page, not here.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.