Skip to main content
Glama

A video in each place's shape

reframe_video
Destructive

Convert one finished video into platform-ready versions for TikTok, Reels, Shorts, YouTube, X and LinkedIn, with motion-following crops, padded shots, length cuts, loudness targets and covers.

Instructions

Use it when the user wants a video for TikTok, Reels, Shorts, YouTube, X or LinkedIn: vertical, square or 4:5, or cut to a place's length or size. Turns one finished video into the version each place takes, with ffmpeg on this machine: 9:16 for TikTok, Reels and Shorts, 4:5 for the feeds, 1:1, and 16:9 for YouTube, X and LinkedIn. In each shot the crop follows what moves, not faces, and a shot whose action or words are wider than the crop is padded, shown whole on a blurred copy of itself. Each file is cut at a shot change to its place's length, kept under its size limit, set to the loudness asked for, given a cover beside it and checked against the limits the place publishes. FORMATS.md, which the result shows, lists every file with its checks; sheet.png, the picture to show first, draws each shape's window on every shot. A launch film's timeline or a demo's steps file beside the video gives its shots. To move a window, plan first and edit crop.json. For a look at the windows before anything renders, use plan_reframe instead. The reframe playbook (makersmcp_playbook) has the whole job. (reframekit 0.1.3, on this machine.)

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "video": "<video>"
    },
    "name": "reframe_video"
  }
}

Input Schema

NameRequiredDescriptionDefault
forNoThe places, in one run at most 12: any of tiktok (1080x1920), reels (1080x1920), shorts (1080x1920), youtube (1920x1080), x (1920x1080), linkedin (1920x1080), feed (1080x1350, for the 4:5 feeds), square (1080x1080), all (every one of them, the default), or a size like 1080x1350, even sides of 64 to 4096 pixels from 1:4 to 4:1.
fpsNoThe files' frame rate, like 30 or 29.97 (default: the video's own, within what the places take).
padNoWhat a padded shot sits on: blur (default, a blurred, darkened copy of itself), black, white, a color like #0f172a, or brand, the page background of the brand kit given as brand.
cropNoA crop.json from plan_reframe or an earlier run, edited by hand: its windows are rendered as they are, and a shape it has none of is planned for its shots. It takes no cuts or steps, and no mode but fit.
cutsNoEvery shot change, in place of those found: seconds like 3.2,7.5, or a JSON file of seconds, of frames with their fps like {"fps": 30, "cuts": [96, 225]}, a launch film's film.timeline.json or a demo's steps.json.
lufsNoThe loudness of every file in LUFS (default -14).
modeNofollow crops each shot where what moves is, and pads a shot a crop would cut (default follow); center crops from the middle; pad shows every shot whole on a blurred copy or a color; fit keeps the picture's own shape and only fits each place's limits.
trimNoSet to false never to cut a file to its place's length, for an account a place allows more, like TikTok's 10 minutes; FORMATS.md then says the file is over.
brandNoA brand kit, its folder, brand.json or _brand.yml: padded shots then sit on its page background color. make_brand_kit makes one.
coverNoThe moment each place's cover shows, in seconds like "4.5" or as "0:04.5" (default: picked).
stepsNoThe steps.json of the demo this video is, when it is not beside the video under its name: it gives the pages, the page box and each click.
videoYesThe finished video, any file ffmpeg reads, like film.mp4 from a launch film or demo.mp4 from a demo. The film.timeline.json or steps.json beside it, under its name, gives its shots.
out_dirNoFolder to save into, absolute or relative to /app. Default: <video-folder>/reframekit-<video-name>, where <video-name> is the name of video without its extension and <video-folder> is the folder video is in. A run replaces the files an earlier run of this kit left there, so give each version its own out_dir.
secondsNoKeep at most this long in each place's video, cut at a shot change: seconds like "30", or "1:30".
wait_secondsNoHow long to wait for the result, counted from the start of the call, before returning a job_id to wait on. Some clients end a call after 60 or 240 seconds; 45 is safe in all of them.

Output Schema

NameRequiredDescriptionDefault
kitYes
docsNoThe notes the text shows, and where each one is.
nextNoWhat to do next, as the Next line says.
toolYes
filesNo
folderNoWhere the kit saved its files.
job_idYes
passedNoFor checks: whether the check passed.
statusYes
continuedNoTrue when this call continued the job of an identical call whose answer did not reach the client, instead of starting another.
exit_codeNo
last_lineNoWhat the kit was doing last.
expires_atNoWhen the download links stop working.
next_actionNo
elapsed_secondsYes

Schema Changelog

Changes observed during successful MCP inspections.

  1. Addedv0.2.0

TDQS

A4.2/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Goes beyond the annotations by disclosing that processing runs locally via ffmpeg rather than remotely, how crops behave ('the crop follows what moves, not faces'), and what artifacts land on disk (FORMATS.md checks, sheet.png). It omits the destructive overwrite of a prior out_dir run from the main text, leaving that to the schema, but the operational profile is otherwise well covered.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded well with the usage trigger and platform list first, but the sentences are long and clause-heavy, packing crop logic, padding, cutting, loudness, cover selection and output files into single sprawling constructs. The content is dense but not easily scannable.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 15-parameter mutation tool, the description covers the engine (local ffmpeg), output artifacts (FORMATS.md, sheet.png), timeline/steps inputs, and the plan-then-edit workflow for crop.json. With an output schema present it needn't explain return values, and the annotations already carry the destructive/idempotency profile.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100% and all 15 parameters are documented inline, so the description only needs to reinforce. It does add conceptual anchors for crop/follow/pad behavior, but most of its phrasing re-states choices the schema already spells out.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a concrete verb and resource ('turns one finished video into the version each place takes') and enumerates the target platforms with their aspect ratios (9:16 for TikTok/Reels/Shorts, 16:9 for YouTube/X/LinkedIn). An agent can distinguish it from plan_reframe because the description explicitly frames that sibling as the planning-only route.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Opens with an explicit when-to-use trigger ('Use it when the user wants a video for TikTok, Reels, Shorts...') and names the alternative with its condition ('For a look at the windows before anything renders, use plan_reframe instead'). It also routes to the playbook for the full workflow.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.