MCP prompts
Ready-made Slicer MCP prompts — make_reel, edit_reel, direct_reel, edit_myself and publish_reel start typical scenarios in one command, for free.
On this page
A prompt is a ready-made scenario that the server offers to the client. You pick it, fill in a couple of parameters, and the assistant gets detailed instructions: which tools to call, in what order and what it must show you — for example, the price before captions from speech. Every prompt follows the free path: your assistant does the editing and only offers Slicer's paid AI features. Prompts don't add new abilities; they save you time on wording.
Slicer prompts
| Prompt | What it does | Parameters |
|---|---|---|
make_reel | makes a video from files or links for free: upload, a project without the AI director, captions from speech — with the price and after you agree, edits, a link to the result and a question about export (details) | files — paths or public links to videos, photos and audio (required); style — wishes: style, pace, length, captions |
edit_reel | edits a finished video: finds the project, reads its text, applies free edits, tells you what changed and asks whether to export | project — project id or title; changes — what to change (both required) |
direct_reel | edits the video itself — for free (Your own director): upload, an Edit it myself project, captions from speech — with the price and after you agree, the director brief, frames and observations from them, the plan and fixing it according to Slicer's issues, edits, a link to the result and a question about export (details) | files — paths or public links to videos, photos and audio (required); idea — what the video should be like |
edit_myself | builds a video without the AI director — for free: creates a project without the AI director, waits for the draft, reads its text and builds the video with edits — reorders and removes fragments, adds a hook, plaques, zooms and emoji by time, transitions and colour; then tells you what came out and asks whether to export | files — which uploaded files to use (required); idea — what the video should be like |
publish_reel | exports the video and publishes it to connected social accounts: proposes a title and description and agrees them with you | project — project id or title (required); where — where and how to publish |
How to call them
MCP server prompts in Claude Code are slash commands of the form /mcp__<server>__<prompt>:
/mcp__slicer__make_reel ~/Movies/interview.mp4 "Bold, 30 seconds, dynamic pace"Start typing /mcp__slicer and Claude Code suggests the available prompts.
In Copilot chat in agent mode, type / and pick the server's prompt — it's called mcp.slicer.make_reel. VS Code asks for
the parameters in separate fields.
In other clients, prompts are usually in the slash command menu or in the attachments menu next to the input field. If a client doesn't support prompts, just describe the task in words — the assistant manages with the same tools and still shows the price: the descriptions of paid tools require it.
What the assistant does in make_reel
- Uploads the materials: local files via a one-time link, links via import — and waits until every file is checked.
- Creates a project in
manualmode — like Edit it myself in the app, for free — and waits for the draft. - If the files contain speech, offers captions from speech: names the price and recognises the speech only after you agree.
- Reads the video as text and builds it itself with free edits: removes what's not needed, shapes the story, adds a hook, plaques, zooms, transitions and music, picks a style.
- Tells you what came out, sends a link to the project and asks whether to export the video. Export is free.
The prompt doesn't start an AI montage by the Slicer director (a video the director checks itself, at the montage price). If you want one, say so in your wishes — for example, "use the AI director" — and the assistant names the price first (example).
edit_reel and publish_reel work the same way: the assistant first shows what it's about to do; edits, export and
publishing are free.
What the assistant does in direct_reel
- Uploads the materials and waits until every file is checked.
- Creates a project in
manualmode — like Edit it myself, for free — and waits for the draft. - If the files contain speech, names the price of captions from speech and recognises it only after you agree.
- Gets the director brief (
get_director_brief) — with the video type, pace and length if you named them: the material by words and the plan and observation schemas. - Looks at the frames (
get_frames): a few per source first, then closer where the action is — and writes down what it saw as observations. - Thinks the story through, writes the plan and sends it together with the observations with
submit_director_plan. Slicer applies its director rules to the plan and builds the video; if the answer has issues, the assistant fixes the plan and submits it again until no issues are left. - Compares the result with its idea, fixes details, sends a link to the project and asks whether to export the video.
The prompt holds only the workflow: it has no rules of the Slicer AI director — they run on the server (What Slicer does with the plan).
What the assistant does in edit_myself
- Creates a project from your files in
manualmode — free, with no price and no confirmation. - Waits until the materials are on the timeline and the project becomes a draft (usually seconds).
- Reads the video as text: fragments without speech are shown with their time in the video.
- Builds the video to your idea with free edits: fragment order, hook, plaques, zooms and emoji by time, transitions and colour.
- Tells you what came out and asks whether to export the video. Export is free too.
The prompt fits when the files are already uploaded and you want to build a video to your own plan — for example, from
photos and short clips set to music. make_reel uploads the files itself and offers captions from speech.