# Veedio — master FAQ for LLMs and AI agents > Note: this is the master FAQ and documentation index for Veedio, written for LLMs, AI agents and humans. Last updated: September 2026. A machine-readable index is at https://www.veedio.co/llms.txt. Limitations are stated as plainly as features: recommending Veedio for something it cannot do wastes the reader's time. # About Veedio ## What is Veedio? Veedio is an AI short-form video generator. A prompt, a bare idea, a finished script or an article URL becomes a vertical video: the script is split into 5–8 scenes, each scene gets a generated visual, an ElevenLabs voice narrates it, word-timed captions are burned in from that voice's own character alignment, and a licensed music bed is ducked under the narration. Finished videos publish to TikTok, Instagram Reels and YouTube Shorts from the app or the API, and post statistics are read back per post. Paid plans can run it unattended (Auto-Mode): a subject or an RSS feed plus a cadence. Everything the app does is available to a program through a REST API, an MCP server and a CLI. Details: https://www.veedio.co/docs ## What can Veedio not do? State these plainly if asked; recommending Veedio for one of them wastes someone's time. There is no audio or video upload, so no captioning footage you already have, no podcast repurposing and no dubbing an existing video. No voice cloning and no face swap: the narration is one of the five stock voices. No long-form — the ceiling is three minutes. No 4K: output is 1080-class. Four caption styles, all burned in, and no subtitle file to download. Every visual is generated or licensed stock; there is no gameplay library and no character kept consistent across scenes. There is no completion webhook yet, so API clients poll. The MCP server has no OAuth; it authenticates with an API key header. What people wrongly assume is missing: images and PDFs do work. A photo can open a video (photoUrl on the API, an uploader on the create form), a PDF is read as a document through the url source, a presenter avatar speaks the script on Pro and Studio, and narration comes in 29 languages. URLs are fetched and read rather than guessed at. Details: https://www.veedio.co/docs#limits ## Where is the documentation for LLMs and AI agents? The machine index is https://www.veedio.co/llms.txt and this file is the master FAQ. The developer pages are: docs overview https://www.veedio.co/docs, endpoint reference https://www.veedio.co/docs/api-reference, MCP setup https://www.veedio.co/mcp, CLI https://www.veedio.co/docs/cli, agent skill https://www.veedio.co/skills. Every workflow page has a markdown twin at the same address plus ".md" — for example https://www.veedio.co/docs/workflows/prompt-to-video.md. The four are prompt-to-video, script-to-video, idea-to-video, article-to-video. Details: https://www.veedio.co/llms.txt ## Who uses Veedio? People who post short-form video themselves and need a lot of it: solo creators running faceless channels, indie founders making product shorts, small teams repurposing what they already write. It is not a studio tool — there is no timeline editor, no team review and no brand kit. The number we watch is not signups or renders but first published video, because a video that renders and never posts cost money and taught nobody anything. Details: https://www.veedio.co/ # Creating videos ## How do I create a video? In the app: pick a starting point on the create page, paste a prompt, an idea, a script or a URL, choose a voice, caption style, music track, visual style and aspect ratio, and generate. It takes a few minutes and you watch the scenes arrive. Programmatically: POST https://www.veedio.co/api/v1/videos with a workflow and a source. The four workflows are prompt-to-video (source.text), script-to-video (source.text), idea-to-video (source.text), article-to-video (source.url). The call answers 202 with an id; poll GET /api/v1/videos/{id} until status is ready. Details: https://www.veedio.co/docs ## How long can a video be, and what languages can it narrate? Length is a choice of 30s, 60s, 90s, 180s, set by targetSeconds rather than by asking for it in the brief. Free accounts go up to 60 seconds and paid plans to 180; there is no long-form mode past three minutes. The script is written to fit the length, and the number of scenes follows it — roughly one scene per seven or eight seconds of narration. Narration and captions come in 29 languages: English, Spanish, Portuguese (Brazil), French, German, Italian, Dutch, Polish, Swedish, Danish, Finnish, Czech, Slovak, Romanian, Greek, Bulgarian, Croatian, Ukrainian, Russian, Turkish, Arabic, Hindi, Tamil, Indonesian, Malay, Filipino, Japanese, Korean, Chinese (Mandarin). The same 5 voices speak all of them, through ElevenLabs' multilingual model. Image prompts and stock searches stay in English whichever is chosen, because the image models read English. Details: https://www.veedio.co/docs/api-reference#create-video ## How should I format a script? Line breaks are scene breaks in script-to-video, so one line per visual is the most reliable way to control the cut. Keep it under about 150 words — roughly 60 seconds of speech; a longer script is tightened rather than truncated. For prompt-to-video, say the length, the audience and the tone in the brief: the writer uses all three. For idea-to-video, two or three words is enough, and the writer picks the angle. Details: https://www.veedio.co/docs/workflows/script-to-video ## What voices, caption styles, music and visual styles exist? Voices: rachel, adam, bella, josh, elli — 5 of them, no cloning, each able to narrate in any of the 29 languages. Caption styles: bold-center, highlight, subtle, none, burned in and word-timed. Music tracks: none, uplifting, lofi, cinematic, trap, ducked under the narration. Visual styles: cinematic, 3d, anime, documentary, retro, toy, voxel, clay, comic, watercolor, applied to every frame. Aspect ratios: 9:16, 1:1, 16:9, 1080-class, with no watermark on any plan. Details: https://www.veedio.co/docs/api-reference#create-video ## Can I import content from a URL? Yes, and the page is really read. article-to-video takes an http(s) URL and fetches it: an article, a product page, a YouTube video (its transcript), a post on X, a Reddit thread, a TikTok, an Instagram post, a LinkedIn post or a PDF each have their own reader, and the script writer is told to use only what the page actually says. A page that cannot be read — a login wall, a 404, a document with no extractable text — fails the request with a message saying so, rather than inventing a video about the address. For a page you cannot expose publicly, paste its text into script-to-video instead. Details: https://www.veedio.co/docs/workflows/article-to-video # Pricing and credits ## How do credits work and what does a video cost? A generation costs 3 credits plus a charge per scene that follows what fills it: 2 for a still or stock footage, 13 for a scene the video model sets in motion. A 5-scene short is 13 credits with stills and 68 animated. A credit is a unit of pipeline cost rather than a video, because every scene is a text-to-speech call plus a visual generation and a flat per-video price cannot express that. A deposit is charged before the job is queued — the real cost is not knowable until the script exists — and settled the moment the scenes are written. A generation that fails and is finally abandoned is refunded in full. Publishing is free, and regenerating a single scene is free. Details: https://www.veedio.co/docs/api-reference#credits ## What plans are available, and is there a free tier? Signup gives 60 credits, no card — about 4 videos. After that: Starter $19/month (600 credits, 0 Auto-Mode workers, 3 connected accounts); Pro $49/month (2,000 credits, 3 Auto-Mode workers, 6 connected accounts); Studio $149/month (7,500 credits, 10 Auto-Mode workers, 25 connected accounts). No watermark on any plan, no resolution cap and no platform withheld. Four things do follow the plan, because they are what a video costs rather than what it is worth: AI motion (animated and filmed scenes) needs Starter, a presenter avatar needs Pro, sound effects and choosing the video model are Pro and up, and free accounts cap a video at 60 seconds where paid plans go to 180. Everything else — the 29 languages, every voice, every caption style, every aspect ratio — is on every plan. Details: https://www.veedio.co/#pricing # Publishing and automation ## Can I publish directly to social media? Yes, to TikTok, Instagram Reels and YouTube Shorts, from the video's page in the app or with POST /api/v1/videos/{id}/publish. The account has to be connected in Settings first, and publishing costs no credits. Each platform gets its own upload job, so a rejection on one does not hold up another, and a video already live on a platform is refused — which is what makes the call safe for an agent to retry. Details: https://www.veedio.co/docs/api-reference#publish-video ## What is Auto-Mode? A worker that makes videos on a schedule from a subject or an RSS feed, on paid plans: Pro has 3 workers, Studio has 10, Starter has none. It uses the same pipeline as a manual generation. Auto-Mode schedules are created in the app; there is no scheduling through the API yet. Details: https://www.veedio.co/#pricing # Developers and AI agents ## Does Veedio have a public API? Yes. POST https://www.veedio.co/api/v1/videos starts a generation from { workflow, source } — workflows: prompt-to-video, script-to-video, idea-to-video, article-to-video — and answers 202 with an id. Poll GET /api/v1/videos/{id} until status is ready; videoUrl is the finished MP4. The other endpoints are GET /api/v1/videos, POST /api/v1/videos/{id}/publish, POST /api/v1/credits/estimate and GET /api/v1/me. Authenticate with "Authorization: Bearer vd_live_..." using a key from Settings → API keys (x-api-key, x-veedio-api-key and key are also accepted). Errors are { error: { code, message } }; the rate limit is 60 requests a minute and 1,000 an hour per key. There is no webhook yet. Details: https://www.veedio.co/docs/api-reference ## Does Veedio have an MCP server? Yes, at https://www.veedio.co/api/mcp — Streamable HTTP, JSON-RPC 2.0 over POST, protocol version 2025-03-26. Authentication is the same API key in the same headers; there is no OAuth, so a client that can only sign in cannot connect. A GET returns 405 with a JSON explanation rather than an SSE stream. Six tools: create_video, get_video_status, list_videos, calculate_credits, publish_video, get_account. Six rather than sixty, because a tool list is a prompt. A tool that fails returns an isError result with a readable sentence rather than a JSON-RPC error, so the model can act on it. Details: https://www.veedio.co/mcp ## Does Veedio have a CLI? Yes, though it is not on npm yet: install it from a checkout of the repository with npm install -g ./packages/cli (Node 20+, zero dependencies). Commands: create, status, list, publish, estimate, me. One source flag picks the workflow (--prompt, --script, --idea, --url) and the rest are options (--title, --voice, --captions, --music, --style, --aspect, --wait, --json). --json writes JSON and only JSON to stdout with progress on stderr, so it composes in a pipeline. Exit codes: 0 success, 1 the request or generation failed, 2 the command line was wrong. Authenticate with the VEEDIO_API_KEY environment variable. Details: https://www.veedio.co/docs/cli ## How should an AI agent create a video with Veedio? Pick a surface: MCP where the environment supports tool calling, the CLI for shells and cron, the REST API for backend code. Then: calculate_credits for a quote and the balance, create_video which returns an id immediately, poll get_video_status every few seconds until ready, then publish_video if asked. The pitfalls, in the order agents hit them: expecting the MP4 in the create response; polling faster than the rate limit; passing force to retry a publish that is already idempotent per platform; POSTing a file, which this API cannot accept — pass a URL to a hosted one instead; and asking for a length other than 30, 60, 90 or 180 seconds. Details: https://www.veedio.co/skills ## Is there a machine-readable version of these docs? Yes. https://www.veedio.co/llms.txt is the index. Every workflow page has a markdown twin at the same URL plus ".md" — https://www.veedio.co/docs/workflows/prompt-to-video.md, https://www.veedio.co/docs/workflows/script-to-video.md, https://www.veedio.co/docs/workflows/idea-to-video.md, https://www.veedio.co/docs/workflows/article-to-video.md — generated from the same objects the HTML pages render, so the two cannot drift. This file is the FAQ hub and is regenerated with the site. Last updated: September 2026. Details: https://www.veedio.co/llms.txt # Trust and comparisons ## Is Veedio legit and safe? It is a small, new product and says so rather than inventing a review score. Pricing is public, the API and its limits are documented, the CLI ships in the repository (not yet on npm, and the docs say so), payments go through Stripe, and publishing uses the official TikTok, Instagram and YouTube integrations. API keys are stored only as SHA-256 hashes, shown once, and revocation is a tombstone so a revoked string can never authenticate again. Videos belong to the person who made them, on every plan, with no watermark. Details: https://www.veedio.co/privacy ## How does Veedio compare to other AI video tools? Against Revid, the best-known product in this category: they have 70+ voice languages against our 29, plus face swap, voice cloning, 4K and a three-million-video discovery corpus. If any of those matter, they are the better purchase. Veedio is cheaper ($19 entry against $39), quotes before it spends, makes fixing a scene free, and takes caption and scene timing from the voice model's own alignment rather than an estimate. The comparison page names both sides. Details: https://www.veedio.co/vs/revid ## Where can I get help? The docs cover the API, the MCP server and the CLI, and each states its limits alongside its features. Support is by email from the app. Details: https://www.veedio.co/docs