# MCP example scenarios > Requests for an AI assistant with the Slicer MCP server — it edits for free; captions from speech, translation and the AI director come with a price. Source: https://docs.slicer.media/en/mcp/examples/ Below are typical conversations with an assistant and the tools it calls along the way. The wording is free-form: the assistant understands requests in any language. Almost every scenario is free — your assistant edits the video, like **Edit it myself** in the app. Where a paid Slicer AI feature is needed, the assistant names the price first. ## A video from a file on your computer Suits assistants that can run commands on your computer. > Make a 30-second reel from `~/Movies/interview.mp4` in the Bold style: remove the pauses, add captions and the hook "A reel in 3 minutes". What the assistant does: 1. `create_upload_link` — creates an upload and a one-time link, then runs `curl -T ~/Movies/interview.mp4 ""`; `get_upload` — waits until the file is checked. 2. `create_project` — creates a project without the AI director, free and with no confirmation; `get_project` — waits for the draft. 3. Captions need recognised speech, so `transcribe_speech` without `confirmCost` — gets the price, tells you the expected and the maximum amount and asks: "Shall I recognise the speech for captions?" 4. After your "yes" — `transcribe_speech` with `confirmCost`; `get_transcribe_jobs` — waits until recognition is done. 5. `get_transcript` — reads the video with its words; `edit_project` — `cut_silences` removes the pauses, `trim_block` and `remove_block` fit the video into 30 seconds, `set_hook` adds the hook; `set_style` — the Bold style. 6. Sends you a link to the project, tells you what came out and, after your "yes", calls `export_video`. Only the captions from speech were paid; the upload, the project, the edits and the export are free. ## A video from a link Suits any client — Slicer downloads the file itself. > Here's the talk recording: [https://media.example.com/talks/launch.mp4](https://media.example.com/talks/launch.mp4). Make a 15-second video from it — the brightest moment about the new feature. The assistant calls `import_from_url`, waits for the upload with `get_upload` and creates a project. To find the right moment it needs to know what's said — so it offers captions from speech with the price, and after recognition keeps the right piece with edits and adds a hook. Without captions the assistant still builds the video, but picks the piece by the time you name. ## A video from photos and clips set to music Free from start to finish: there's no speech, so captions from speech aren't needed. > Make a 9:16 draft from `morning-1.mp4`, `morning-2.mp4`, `latte.jpg` and `track.mp3`. Start with the photo, add the hook "Morning at the coffee shop", a title "Cappuccino in 3 minutes" over the second clip and soft transitions, with cuts on the beat. Then export it. What the assistant does: 1. Uploads the files, as in the first scenario. 2. `create_project` — no price and no confirmation; the track goes onto the music lane by itself. `get_project` — waits for the `draft` status. 3. `get_transcript` — sees the fragments in upload order with their time in the video. 4. `edit_project` — `move_block` (the photo to the start), `set_hook`, `add_title` by time (`start`, `duration`), `set_all_transitions` with a fade, `fit_music` and `fit_to_beats`. 5. Tells you what came out and, after your "yes", calls `export_video`. The same in one command — the [`edit_myself`](https://docs.slicer.media/en/mcp/prompts/#edit-myself) prompt. ## Captions for an existing draft > Add captions to the "Morning at the coffee shop" video and highlight the drink names. Tell me the price first. `list_projects` → `transcribe_speech` without `confirmCost` (the price: expected and maximum) → your consent → `transcribe_speech` with `confirmCost` → `get_transcribe_jobs` until it's ready → `get_transcript` → `edit_project` with `set_accent`. You can also learn the price in advance — `estimate_cost` with `kind: "transcribe"` and `projectId`. If recognition fails, the tokens come back. ## Editing text and design > In the "New collection overview" video, highlight the color names, put a ✨ emoji at the start and add a list plaque "Three looks for the week". `get_transcript` → `edit_project` with `set_accent`, `set_emoji` and `add_plaque` operations. Free; the assistant tells you if some edit couldn't be applied and why. > Make all fragments look like the first one — the same filter and transition — and use the frame at 0:05 as the cover, with the title "New collection". `get_transcript` → `edit_project` with `apply_look` and `set_cover` operations. Free. > Make the captions bigger, the font poster, the color amber. `set_style` — free; the new style appears in the file after export. > Make the spoken word pop in the captions. `set_style` with `subtitles: { highlight: "pop" }` — free: the spoken word grows and lights up while the others make room. The design, size and other caption settings stay the same. Requests like "Let the words appear as they are spoken" (`animation: "words"`), "Show one word at a time" (`words: "one"`) and "Use the Box design" (`design: "box"`) work the same way — [all the fields](https://docs.slicer.media/en/api/editing/#style-subtitles). ## Export and publishing > Export the video in 1080p and publish it to YouTube Shorts and Instagram Reels. Write the description yourself, hashtags — #editing #reels. `list_social_accounts` → `export_video` with a profile and a list of publications → `get_project` until rendering is done. Free. The assistant shows you the description before publishing. TikTok isn't available over MCP — its rules require publishing from the Slicer app. ## A batch from a folder > Make a separate video in the Clean style from each video in `~/Movies/lessons`, with a hook from the lesson title, and export them. The assistant uploads the files, creates a project for each one, sets the style and the hook and exports — all for free. If you want captions from speech, it names the total recognition price first. For large batches, a script is more reliable — see the [Batch processing](https://docs.slicer.media/en/api/recipes/batch-python/) recipe. ## The assistant edits by itself, like a director The assistant can edit the video by itself and for free: it looks at the frames and writes the plan, and Slicer applies the rules of its AI director to the plan on the server and answers with what it corrected itself and what the assistant should fix. > Edit a video yourself in Slicer from `~/Movies/latte-art.mp4` and `~/Movies/cafe.mp4`: 30 seconds, dynamic pace, hook "Latte art in a minute". What the assistant does: 1. Uploads the files and creates a project without the AI director, as in the first scenario. 2. If there's speech — names the price of captions from speech and, after your yes, recognises it (`transcribe_speech`). 3. `get_director_brief` with `pace: "dynamic"` and `targetDurationSec: 30` — the material by words and the plan and observation schemas. 4. `get_frames` — 8 frames per video first, then closer around the main action; it writes down what it saw as observations. 5. `submit_director_plan` — the plan and the observations; Slicer builds the video and answers with fixes and issues, the assistant fixes the plan accordingly and submits it again. 6. Fixes details with `edit_project`, sends a link to the project and, after your yes, runs `export_video`. Only speech recognition was paid. In detail, with the `direct_reel` prompt — [Your own director](https://docs.slicer.media/en/mcp/own-director/#example). ## AI montage by the Slicer director If you want the video edited by the Slicer AI director rather than by your assistant, ask for it directly. The director picks the best moments, places captions, emoji, plaques and camera moves itself, then reviews and fixes the finished video. It's paid: 24 tokens per video up to 90 seconds; the first video is free. > Have the Slicer AI director edit the interview `~/Movies/interview.mp4`: Bold, 30 seconds, dynamic pace. What the assistant does: 1. Uploads the file, as in the first scenario. 2. `create_project` with `mode: "ai"` without `confirmCost` — gets the price and asks you: "An AI director montage costs 24 tokens. Shall I start?" 3. After your "yes" — the same call with `confirmCost`; `get_project` with waiting — follows the montage (usually 3–6 minutes). 4. Shows you the finished video and tells you what's in it. If you want the video from before a rebuild back — `choose_version`. Rebuilding with other settings (`remount_project`) works the same way, at the montage price and after you agree. Retrying a failed run (`retry_project`) costs the same as the original run; a free run is retried for free. ## Caption translation > Translate the video's captions into Spanish. Translation is a paid AI feature at cost: the assistant names the expected and the maximum price (`translate_captions` without `confirmCost`) and translates after you agree; the actual price is charged, never more than the maximum. The translation is laid out word by word, so accents and emoji stay in place; the new text shows up in the file after export. ## Questions about Slicer > How do I make a freeze frame in Slicer? And how much does background removal cost? `search_docs` → `read_doc` — the assistant answers from these docs with a link to the page. Free. > **Tip.** No paid feature starts without your consent to the price. Other ways to limit spending — the connection's spend limit or scopes — are on the [Security and costs](https://docs.slicer.media/en/mcp/costs/#scopes) page. ## What's next [ToolsEvery tool and its parameters.](https://docs.slicer.media/en/mcp/tools/)[LimitationsWhat the assistant can't do yet.](https://docs.slicer.media/en/mcp/limits/)