inReels

Erstelle Videoanzeigen, UGC-Anzeigen, Bilder, Voiceovers und gesichtslose Videos mit deinem KI-Agenten. Remote-MCP mit OAuth.

Dokumentation

Make ads and videos from Claude, ChatGPT or Cursor

Paste one URL into your AI app, then ask in plain words.

Paste this URL into Claude, ChatGPT or Cursor as a connector

Add it as a custom connector (MCP) and sign in to inReels when it asks. No key needed.

https://www.inreels.ai/api/mcp

Or hand it to your agent

Copy the setup and paste it in your agent's chat, or just give it this link. It will set inReels up and guide you.

https://www.inreels.ai/connect.md

Steps for each app (and the API)

  1. 1 In Claude, open Settings → Connectors → Add custom connector.
  2. 2 Name it inReels, paste this URL, then Connect and sign in.
    https://www.inreels.ai/api/mcp
    
    Copy
  3. 3 In a chat, turn inReels on from the + menu and paste the first prompt below.

Then try this first

Use inReels to make a 15-second UGC video ad for https://your-site.com. Show me the price before you start.

It quotes the price first and only spends credits from your inReels account. Failed runs are refunded.

21 tools · 41 image, video, speech and sound models · OAuth or API key

Developer reference: tools, behaviour, errors⌄

What the agent gets

Capabilities — a finished deliverable

These choose the model, apply our craft rules and return a file. Creative controls only.

create_ad_video sensitive writespends credits

A finished ad from a product image, with native audio. Add a creator (actor_id or image) and it becomes a UGC ad; leave it out for a product-led one. Supply a script and it sets the length — write to the budget: up to 10 words for 5s, 20 for 10s, 30 for 15s. Over budget and the line gets clipped, so cut words rather than hoping. Omit the script and pass a brief instead, and the model writes its own line to fit. Ask for up to 3 variations to test angles. The model is chosen for you; use generate_video to pick one yourself. Needs a paid plan. Docs: https://www.inreels.ai/connect

create_faceless_video sensitive writespends credits

Make a complete faceless story video (9:16, about 30-60s): from a topic or a script it writes the story, paints the shots in the look the story calls for (hand-painted by default; anime, blocky voxel, stylized 3D or realistic cinematic when the topic or visual_style names one), narrates, scores and renders a finished MP4, and returns a run id at once (status planning). It takes several minutes; poll get_run_status (the title appears there once written) until it reports completed, and only then is output_url an MP4. Credits are charged once when making starts and refunded in full if it fails. The exact price is set once the story is written: up to 25% over the quote it goes ahead; beyond that nothing is charged and get_run_status reports phase price_changed with new_price. The look, voice, music and length are chosen by the pipeline — voice_id, visual_style, music_id, niche, image_quality, scene_count and duration are accepted for compatibility but not used (the result lists any you set in notes).

create_video sensitive writespends credits

Make a finished, multi-scene video in one of our formats and return a run id — not a single model clip (that is generate_video). Poll get_run_status until it reports completed; it takes several minutes. Pick format from list_formats when you know which fits. LEAVE FORMAT OUT when you are not sure: a plain video is then made from your brief, rather than us guessing a specialist format and spending your credits on the wrong thing. Set max_credits to cap the spend — if the real price lands above it nothing is started and nothing is charged, and the error gives the real figure. Anything you pass that the chosen format has no use for is named back in notes, never dropped in silence.

Primitives — one asset, your choice of model

Full parameter surface across 41 image, video, speech and sound models.

generate_image sensitive writespends credits

Render an image and return a run id. Poll get_run_status for the URL. Call quote first if the budget matters.

generate_video sensitive writespends credits

Render a clip and return a run id. Text-to-video, or image-to-video by passing image_url. Several models produce native audio — set generate_audio and check list_models for which support it. Takes minutes; poll get_run_status.

generate_sound sensitive writespends credits

Make a sound effect or a music bed and return a run id. Poll get_run_status for the audio URL and `duration_seconds`. Effects run 1-22 seconds, music 10-300 and instrumental by default so it sits under a voiceover. This is the sound half of a video: without it a cut lands silent, and a bed under narration is what makes a montage feel scored rather than assembled.

generate_speech sensitive writespends credits

Speak a line in a named voice and return a run id. Poll get_run_status for the audio URL — it also returns `words`, the start and end second of every word, and `duration_seconds`. Use those to cut to the voice or build word-accurate captions instead of running speech-to-text over audio you just made.

Discovery, pricing and runs

get_account readfree, read only

Confirm the connection works and read the credit balance. Call this first if anything is failing.

list_models readfree, read only

Every image, video, speech and sound model, with what each accepts and its price. Read this before naming a model — an id that is not listed is rejected.

list_voices readfree, read only

Named voices available to generate_speech.

quote readfree, read only

Price a generate_image, generate_video, generate_speech or generate_sound call without spending anything. The number returned is exactly what the matching call will charge.

get_run_status readfree, read only

Check a submitted run. Pass wait:true to block up to 45 seconds, which usually returns a finished image in one call. Always poll before giving a user an output URL. Video runs keep going after the scenes are made — rendering the MP4 takes a few more minutes, so keep polling until status is completed and tell the user it is rendering rather than handing over the watch_url as the finished thing.

list_runs readfree, read only

The most recent runs on this account, newest first.

list_styles readfree, read only

The valid values for create_faceless_video. Read this before submitting — an unlisted style or music id is rejected, not substituted.

list_actors readfree, read only

The curated faces available for create_ad_video, with filters. Pick deliberately — the actor is most of what makes a UGC ad read as real. To use a face of your own instead, call upload_image and pass its URL.

upload_image writefree

Bring your own product shot or creator photo in, by URL or base64, and get back an inReels URL the generation tools accept. Free. PNG, JPEG or WebP, up to 10MB. Docs: https://www.inreels.ai/connect

refresh_asset_url readfree, read only

Output and upload URLs are signed and expire after 24 hours. Pass the s3_key to get a fresh link for the same file.

quote_ad_video readfree, read only

Price an ad without making it. Returns the model chosen, the duration the script implies, and the total for all variations.

quote_faceless_video readfree, read only

Estimate a faceless story video without submitting it. Free plans get still painted shots; paid plans also get moving shots, whose exact price is set once the shot list is written, so on a paid plan this is a floor.

list_formats readfree, read only

The kinds of finished video inReels makes, with what each one looks like, what it needs from you, and roughly what it costs. Read this before create_video if you are not sure which format fits — picking the wrong one spends credits on the wrong thing. If none of them clearly fits, leave format out of create_video and a plain video is made from your brief.

quote_video readfree, read only

Price a video in any format without making it. Free, and holds nothing: there is no quote to expire or pass on — create_video re-prices the same brief and max_credits is what protects you from a surprise. A format with a free planning step is priced after it, so the figure is exact.

How it behaves

  • Every generating tool returns a run id immediately. Nothing blocks.
  • Quote first — the price you are shown is the price you are charged.
  • Failed runs are refunded in full, so a failure never costs credits.
  • An unlisted model is rejected, never quietly substituted.
  • Video keeps going after the scenes are made: the MP4 is rendered next, and the run only reports completed once the file exists.
  • Every option we expose is validated against the provider's live schema before it ships, so an agent cannot be handed a combination that fails.
  • No rate limits beyond the account's credit balance. Any number of runs may be in flight; get_run_status with wait blocks up to 45 seconds.

Tool classification

Each tool above is tagged read, write or sensitive write.

  • read — retrieves information. Changes nothing, spends nothing.
  • write — creates or changes data. Spends nothing.
  • sensitive write — spends the user's credits. Irreversible, so a host should ask the user each time rather than taking blanket permission.

Errors

400  invalid_request        rejected before any charge; the message names the fix
401  unauthorized           WWW-Authenticate points at the OAuth metadata
402  insufficient_credits   includes required and available
403  subscription           the capability needs a paid plan
404  not_found              unknown tool, or a run on another account
405  method_not_allowed     a spending tool was called with GET
500  internal_error         failed on our side; nothing was charged

Open source

Both repos are MIT. Nothing here is a black box, and nothing locks you in — the skills run on your own Replicate, fal or ElevenLabs keys just as happily as on ours.

[

inreels-mcp

The stdio server and CLI, for clients without remote MCP or to keep the key in your own environment.

](https://github.com/inreels-ai/inreels-mcp)[

inreels-skills

Agent skills for UGC ads, static creatives, faceless shorts and animated film — the method and the render scripts, not just the API call.

](https://github.com/inreels-ai/inreels-skills)