Turn long videos into captioned vertical shorts, transcripts and articles with MnogoReels.
ru.mnogo-reels/mnogoreels (MCP Server)
The ru.mnogo-reels/mnogoreels MCP server is described as a tool that turns long videos into captioned vertical shorts, along with transcripts and articles. It is positioned as “MnogoReels,” based on the provided server description.
🛠️ Key Features
Captioned vertical shorts
Transcripts
Articles
Video-to-short conversion for long videos
🚀 Use Cases
Converting long-form video into vertical, captioned short clips
Producing transcripts from long videos
Generating article-style outputs from video content
⚡ Developer Benefits
Standardized server access (MCP) for the stated conversion outputs: shorts, transcripts, and articles
Clear input-to-output categories aligned to long-video processing
⚠️ Limitations
Only the capabilities explicitly listed in the server description are supported here; no additional features, tool count, or topics are provided in the source data.
Captured live from the server via tools/list.
get_account
Checks whether the configured MnogoReels API key works and shows this server's settings. Read-only and free; works without a key.
Use it first when other tools fail with an authentication error, or when the user asks which account or mode is active. It does not change anything.
Returns authenticated (true or false), a masked key hint (last 4 characters only), the API URL, the server version and transport, the active profile, enabled toolsets, whether read-only mode is on, the per-call credit cap if one is set, and links for managing keys and credits. The credit balance itself is not available through the API; the user can see it on mnogo-reels.ru.
Explains the option values accepted by create_clip_job and recut_clip_job: video modes, clip lengths, caption styles, hook styles, story arc styles, outros, framing and languages, plus the pricing rules and the 30-minute limit per job. Read-only and free; works without an API key and makes no network calls.
Use it when the user asks what a style looks like, which mode fits their video, or how much processing costs. topic selects one section; all returns everything.
Checks whether MnogoReels can download a video link and returns its title, length and platform. Free: spends no credits and creates no job.
Use it to answer "is this link usable?" or "how long is this video?" without pricing anything (private, removed, region-locked, audio-only or very large sources fail). To price a job call quote_clip_job directly: it runs the same check. Do not use it for existing jobs; use get_clip_job instead.
Supports YouTube, VK Video, RuTube, Dzen, Twitch, Yandex Disk and Google Drive file links, and direct links to .mp4, .mov, .mkv or .webm files. Local files are not supported: the user must upload the file somewhere public first.
Returns downloadable (true or false), reason when not downloadable (audio_only, too_large, unavailable), duration_sec (null when the platform does not report it, which is typical for direct file links), title, thumbnail_url and platform. Usually takes 2 to 15 seconds, occasionally up to 45.
Parameters1
source_url
string
required
Public link to the video: YouTube, VK Video, RuTube, Dzen, Twitch, Yandex Disk file, Google Drive file, or a direct .mp4/.mov/.mkv/.webm link. Local files are not supported.
Raw schema
{
"type": "object",
"properties": {
"source_url": {
"type": "string",
"minLength": 8,
"maxLength": 2083,
"description": "Public link to the video: YouTube, VK Video, RuTube, Dzen, Twitch, Yandex Disk file, Google Drive file, or a direct .mp4/.mov/.mkv/.webm link. Local files are not supported."
}
},
"required": [
"source_url"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
quote_clip_job
Calculates the credit cost of create_clip_job or recut_clip_job without spending anything, and checks that MnogoReels can download the video. Free and read-only.
Price: 1 credit per started minute of the processed range (end_sec minus start_sec), minimum 1. Options such as caption style, hooks, B-roll or music do not change the price. One job processes at most 30 minutes; for longer videos quote several ranges with start_sec and end_sec. Without start_sec and end_sec the whole video is priced, so its length must be known; this tool looks it up.
Pass source_url to price a new video, or job_id to price a recut of an existing clip job. A recut without start_sec and end_sec reuses the original job's range, which the API does not reveal; the quote is then an upper bound (the whole source video) and credits_exact is false. Show the credits to the user and get their agreement before calling create_clip_job or recut_clip_job with max_credits of at least this amount.
Returns credits, credits_exact, billed_seconds, the resolved start_sec and end_sec, the platform, whether the video can be downloaded, the source title and length, and a note about failed jobs. If the length cannot be determined (typical for direct file links), the error explains how to pass end_sec yourself.
Parameters4
source_url
string
optional
Public link to the video: YouTube, VK Video, RuTube, Dzen, Twitch, Yandex Disk file, Google Drive file, or a direct .mp4/.mov/.mkv/.webm link. Local files are not supported.
job_id
string
optional
Existing clip job to price a recut for. Give either source_url or job_id.
start_sec
number
optional
Start of the part to process, in seconds from the beginning. Default 0.
end_sec
number
optional
End of the part to process, in seconds. Default: the end of the video (its length is looked up automatically when the platform reports it). At most 30 minutes after start_sec.
Raw schema
{
"type": "object",
"properties": {
"source_url": {
"type": "string",
"minLength": 8,
"maxLength": 2083,
"description": "Public link to the video: YouTube, VK Video, RuTube, Dzen, Twitch, Yandex Disk file, Google Drive file, or a direct .mp4/.mov/.mkv/.webm link. Local files are not supported."
},
"job_id": {
"description": "Existing clip job to price a recut for. Give either source_url or job_id.",
"type": "string",
"pattern": "^[0-9a-f]{32}$"
},
"start_sec": {
"description": "Start of the part to process, in seconds from the beginning. Default 0.",
"type": "number",
"minimum": 0,
"maximum": 86400
},
"end_sec": {
"description": "End of the part to process, in seconds. Default: the end of the video (its length is looked up automatically when the platform reports it). At most 30 minutes after start_sec.",
"type": "number",
"minimum": 1,
"maximum": 86400
}
},
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
create_clip_job
Starts AI clipping of a long video into vertical 9:16 short clips with burned-in captions and a hook title, plus optional cold-open hooks, music and B-roll. Spends credits: 1 per started minute of the processed range (minimum 1), charged when the job is created. One job processes at most 30 minutes of video; split longer videos with start_sec and end_sec. Jobs cannot be cancelled.
Before calling, run quote_clip_job, show the price to the user and get explicit agreement. max_credits is required; the call is refused without charging if the price is higher. Some clients also ask the user to confirm. If the same request was already submitted with this API key in the last 10 minutes, that job is returned instead of charging again.
Use for turning a podcast, stream, interview, lecture or vlog into Shorts, Reels or TikTok clips. Not for: transcripts only (create_transcription), changing the look of an existing job (recut_clip_job), or re-running a failed job for free (retry_clip_job).
Key parameters: source_url (public link; local files are not supported), start_sec and end_sec to process only part of the video, video_mode (podcast for talking heads), clip_length, caption_style, language (omit to auto-detect) and prompt to steer which moments are picked. Call get_options_guide to explain option values.
Returns job_id right away with status queued and the credits charged. Processing takes a few minutes for short videos and up to tens of minutes for a full 30-minute range; clips appear one by one. Next, call get_clip_job with wait_sec=50 until the status is completed or failed. If processing fails, credits are not returned automatically: retry_clip_job re-runs the job for free, or the user can contact support@mnogo-reels.ru with the job_id. On the free plan clip files are kept for 30 days.
Parameters23
source_url
string
required
Public link to the video: YouTube, VK Video, RuTube, Dzen, Twitch, Yandex Disk file, Google Drive file, or a direct .mp4/.mov/.mkv/.webm link. Local files are not supported.
start_sec
number
optional
Start of the part to process, in seconds from the beginning. Default 0.
end_sec
number
optional
End of the part to process, in seconds. Default: the end of the video (its length is looked up automatically when the platform reports it). At most 30 minutes after start_sec.
max_credits
integer
required
Spending limit for this call in credits. Required. Use the credits value from quote_clip_job after the user agreed; the call is refused without charging if the price (or its upper bound) is higher.
video_mode
string
optional
How the video was filmed: general, podcast (talking heads), general-beta (SmartCut: removes pauses and filler words), stream (screen + webcam), vlog, fitness.
clip_length
string
optional
Target length of each clip. auto lets the AI decide; 30_60s suits Shorts, Reels and TikTok.
caption_style
string
optional
Burned-in caption style; empty means no captions. Call get_options_guide to see what each style looks like.
hook_style
string
optional
Style of the hook title shown at the start of each clip; none hides it.
language
string
optional
Two-letter language code of the speech, e.g. en or ru. Omit to auto-detect. Never pass 'auto'.
prompt
string
optional
What kind of moments to pick, in plain words, e.g. 'funny moments about money'.
cold_open
boolean
optional
Start each clip with its strongest moment as a teaser, and also produce standalone hook clips and a looped teaser.
story_arc_style
string
optional
Build each clip as a story (cold open, main part, payoff). Any value except none also turns on cold_open.
add_music
boolean
optional
Add quiet background music from the MnogoReels library.
emojis
boolean
optional
Add emojis to captions where they fit.
smart_cut
boolean
optional
Remove pauses and filler words inside clips. Works only with video_mode podcast (general-beta always does it); ignored for other modes.
b_roll_stock_video
boolean
optional
Insert relevant stock video B-roll.
b_roll_ai_images
boolean
optional
Insert AI-generated illustrative images as B-roll.
b_roll_motion_graphics
boolean
optional
Insert AI motion-graphics B-roll (animated titles, charts, diagrams).
outro
string
optional
Ending added to every clip: default branded outro, soon/like short Russian text cards, custom = your saved outro image.
framing
string
optional
auto = smart crop; full_frame = keep the whole wide frame on a blurred background; portrait_only = always crop to the subject.
keep_whole_video
boolean
optional
Do not cut into clips: render the whole video once with captions, music and B-roll. Use only for videos that are already vertical and short.
agentic_mode
boolean
optional
Extra AI review pass that refines where clips start and end. Slower.
allow_duplicate
boolean
optional
Set true only if the user explicitly wants a second identical job within 10 minutes (it is charged again).
Raw schema
{
"type": "object",
"properties": {
"source_url": {
"type": "string",
"minLength": 8,
"maxLength": 2083,
"description": "Public link to the video: YouTube, VK Video, RuTube, Dzen, Twitch, Yandex Disk file, Google Drive file, or a direct .mp4/.mov/.mkv/.webm link. Local files are not supported."
},
"start_sec": {
"description": "Start of the part to process, in seconds from the beginning. Default 0.",
"type": "number",
"minimum": 0,
"maximum": 86400
},
"end_sec": {
"description": "End of the part to process, in seconds. Default: the end of the video (its length is looked up automatically when the platform reports it). At most 30 minutes after start_sec.",
"type": "number",
"minimum": 1,
"maximum": 86400
},
"max_credits": {
"type": "integer",
"minimum": 1,
"maximum": 100000,
"description": "Spending limit for this call in credits. Required. Use the credits value from quote_clip_job after the user agreed; the call is refused without charging if the price (or its upper bound) is higher."
},
"video_mode": {
"default": "general",
"description": "How the video was filmed: general, podcast (talking heads), general-beta (SmartCut: removes pauses and filler words), stream (screen + webcam), vlog, fitness.",
"type": "string",
"enum": [
"general",
"podcast",
"general-beta",
"stream",
"vlog",
"fitness"
]
},
"clip_length": {
"default": "30_60s",
"description": "Target length of each clip. auto lets the AI decide; 30_60s suits Shorts, Reels and TikTok.",
"type": "string",
"enum": [
"auto",
"up_to_30s",
"30_60s",
"60_90s",
"90s_3m",
"3m_5m",
"5m_10m"
]
},
"caption_style": {
"default": "karaoke",
"description": "Burned-in caption style; empty means no captions. Call get_options_guide to see what each style looks like.",
"type": "string",
"enum": [
"karaoke",
"karaoke_static",
"minimal",
"viral",
"montserrat",
"gilroy",
"rubik",
"bebas",
"impact",
"pixel",
"gothic",
"going",
"hormozi",
"hormozi_pop",
"boxed",
"boxed_black",
"boxed_pop",
"neon",
"pop3d",
"tiktok",
"money",
"chunky",
"punch",
"empty"
]
},
"hook_style": {
"default": "default",
"description": "Style of the hook title shown at the start of each clip; none hides it.",
"type": "string",
"enum": [
"default",
"dark",
"accent",
"sticker",
"marker",
"neon",
"stagger",
"torn",
"keyword",
"impact",
"none"
]
},
"language": {
"description": "Two-letter language code of the speech, e.g. en or ru. Omit to auto-detect. Never pass 'auto'.",
"type": "string",
"pattern": "^[a-z]{2}$"
},
"prompt": {
"description": "What kind of moments to pick, in plain words, e.g. 'funny moments about money'.",
"type": "string",
"minLength": 1,
"maxLength": 1000
},
"cold_open": {
"default": false,
"description": "Start each clip with its strongest moment as a teaser, and also produce standalone hook clips and a looped teaser.",
"type": "boolean"
},
"story_arc_style": {
"default": "none",
"description": "Build each clip as a story (cold open, main part, payoff). Any value except none also turns on cold_open.",
"type": "string",
"enum": [
"none",
"subtle",
"dynamic",
"cinematic"
]
},
"add_music": {
"default": false,
"description": "Add quiet background music from the MnogoReels library.",
"type": "boolean"
},
"emojis": {
"default": false,
"description": "Add emojis to captions where they fit.",
"type": "boolean"
},
"smart_cut": {
"default": false,
"description": "Remove pauses and filler words inside clips. Works only with video_mode podcast (general-beta always does it); ignored for other modes.",
"type": "boolean"
},
"b_roll_stock_video": {
"default": false,
"description": "Insert relevant stock video B-roll.",
"type": "boolean"
},
"b_roll_ai_images": {
"description": "Insert AI-generated illustrative images as B-roll.",
"type": "boolean"
},
"b_roll_motion_graphics": {
"description": "Insert AI motion-graphics B-roll (animated titles, charts, diagrams).",
"type": "boolean"
},
"outro": {
"default": "none",
"description": "Ending added to every clip: default branded outro, soon/like short Russian text cards, custom = your saved outro image.",
"type": "string",
"enum": [
"none",
"default",
"soon",
"like",
"custom"
]
},
"framing": {
"default": "auto",
"description": "auto = smart crop; full_frame = keep the whole wide frame on a blurred background; portrait_only = always crop to the subject.",
"type": "string",
"enum": [
"auto",
"full_frame",
"portrait_only"
]
},
"keep_whole_video": {
"default": false,
"description": "Do not cut into clips: render the whole video once with captions, music and B-roll. Use only for videos that are already vertical and short.",
"type": "boolean"
},
"agentic_mode": {
"default": false,
"description": "Extra AI review pass that refines where clips start and end. Slower.",
"type": "boolean"
},
"allow_duplicate": {
"default": false,
"description": "Set true only if the user explicitly wants a second identical job within 10 minutes (it is charged again).",
"type": "boolean"
}
},
"required": [
"source_url",
"max_credits"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
get_clip_job
Returns the status, progress and results of a clip job created by create_clip_job, recut_clip_job or on the MnogoReels website. Read-only and free.
wait_sec (default 30, max 50) waits inside this call: it returns as soon as the job completes or fails, or when wait_sec runs out, and sends progress notifications meanwhile. wait_sec=0 returns an instant snapshot. While the job is queued or processing only the count of ready clips is returned; set include_partial_clips=true to get the links of clips finished so far. If eta_seconds is above 600, tell the user how long it will take and offer to check back later instead of polling continuously.
Once completed, returns each clip's video link, title, virality score (0 to 100), duration and hook text, plus cold-open hook clips and teasers when the job made them. Clips are sorted best score first; max_clips limits how many are returned (default 10, up to 30; up to 10 with response_format=detailed). response_format=detailed adds clip descriptions, transcript excerpts, poster images and timings in the source video.
Notes: a failed job can return to queued when MnogoReels retries it automatically. media_available=false means the files were deleted by storage retention (free-plan clips are kept for 30 days). Text from the video is wrapped in <untrusted_source_text> tags. Use list_clip_jobs to find job ids.
Parameters5
job_id
string
required
Clip job id (32 hex characters) returned by create_clip_job, recut_clip_job or list_clip_jobs.
wait_sec
integer
optional
Seconds to wait inside this call for the job to finish (default 30, max 50; 0 = return a snapshot immediately). Progress notifications are sent while waiting.
response_format
string
optional
concise: links, titles, scores. detailed: adds descriptions, transcript excerpts, posters and timings (max_clips up to 10).
max_clips
integer
optional
Maximum number of clips to return, best first (up to 30; up to 10 with detailed).
include_partial_clips
boolean
optional
true returns the links of clips finished so far while the job is still processing.
Raw schema
{
"type": "object",
"properties": {
"job_id": {
"type": "string",
"pattern": "^[0-9a-f]{32}$",
"description": "Clip job id (32 hex characters) returned by create_clip_job, recut_clip_job or list_clip_jobs."
},
"wait_sec": {
"default": 30,
"description": "Seconds to wait inside this call for the job to finish (default 30, max 50; 0 = return a snapshot immediately). Progress notifications are sent while waiting.",
"type": "integer",
"minimum": 0,
"maximum": 50
},
"response_format": {
"default": "concise",
"description": "concise: links, titles, scores. detailed: adds descriptions, transcript excerpts, posters and timings (max_clips up to 10).",
"type": "string",
"enum": [
"concise",
"detailed"
]
},
"max_clips": {
"default": 10,
"description": "Maximum number of clips to return, best first (up to 30; up to 10 with detailed).",
"type": "integer",
"minimum": 1,
"maximum": 30
},
"include_partial_clips": {
"default": false,
"description": "true returns the links of clips finished so far while the job is still processing.",
"type": "boolean"
}
},
"required": [
"job_id"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
list_clip_jobs
Lists the user's clip jobs, newest first, including jobs started on the MnogoReels website. Read-only and free.
Use it to find a job_id, see what is still processing, or pick recent results to review. Use get_clip_job for the clips of one job.
created_via=api shows only jobs started with an API key. Pages with limit (1 to 50, default 10) and offset; next_offset is null on the last page. Creation dates are not provided by the API; the order is newest first.
Returns for each job: job_id, status, progress_percent, source_url, the number of clips ready, pack_id when the job belongs to a publication pack, likely_kind (clip_job, pack_clip_job or republication) and a short error for failed jobs.
Parameters3
limit
integer
optional
Items per page, 1–50 (default 10).
offset
integer
optional
Items to skip; pass next_offset from the previous page.
created_via
string
optional
api = only jobs created with an API key (this server or scripts); any = also jobs from the website.
Raw schema
{
"type": "object",
"properties": {
"limit": {
"default": 10,
"description": "Items per page, 1–50 (default 10).",
"type": "integer",
"minimum": 1,
"maximum": 50
},
"offset": {
"default": 0,
"description": "Items to skip; pass next_offset from the previous page.",
"type": "integer",
"minimum": 0,
"maximum": 100000
},
"created_via": {
"default": "any",
"description": "api = only jobs created with an API key (this server or scripts); any = also jobs from the website.",
"type": "string",
"enum": [
"any",
"api"
]
}
},
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
recut_clip_job
Creates a new clip job from the same source video as an existing job, with different settings such as caption style, hook style, clip length, language, framing, music, B-roll or prompt. Spends credits like create_clip_job (1 per started minute of the billed range, minimum 1, at most 30 minutes per job) and returns a new job_id; the original job and its clips stay unchanged.
Before calling, run quote_clip_job with job_id, show the price, get the user's agreement and pass max_credits. Identical requests within 10 minutes return the existing recut instead of charging again.
Without start_sec and end_sec the recut reuses the original job's range and is billed for it; the API does not reveal that range, so the quote is an upper bound (the whole source video, which must then be at most 30 minutes long). By default the same moments are re-rendered with the new look, which is faster because the source is cached; this works only when the range equals the original job's range. Set new_moments=true to let the AI pick different moments. Only the settings you pass change; everything else (including the original outro and whole-video mode) is inherited from the original job.
Works for completed and failed clip jobs, not for jobs that are still running. To re-run a failed job with the same settings for free, use retry_clip_job. Next, call get_clip_job with the new job_id and wait_sec=50.
Parameters23
job_id
string
required
Clip job to recut.
max_credits
integer
required
Spending limit for this call in credits. Required. Use the credits value from quote_clip_job after the user agreed; the call is refused without charging if the price (or its upper bound) is higher.
start_sec
number
optional
Start of the part to process, in seconds from the beginning. Default 0.
end_sec
number
optional
End of the part to process, in seconds. Default: the end of the video (its length is looked up automatically when the platform reports it). At most 30 minutes after start_sec.
new_moments
boolean
optional
true = the AI picks different moments; false = same moments with the new look.
allow_duplicate
boolean
optional
Set true only if the user explicitly wants a second identical job within 10 minutes (it is charged again).
video_mode
string
optional
How the video was filmed: general, podcast (talking heads), general-beta (SmartCut: removes pauses and filler words), stream (screen + webcam), vlog, fitness.
clip_length
string
optional
Target length of each clip. auto lets the AI decide; 30_60s suits Shorts, Reels and TikTok.
caption_style
string
optional
Burned-in caption style; empty means no captions. Call get_options_guide to see what each style looks like.
hook_style
string
optional
Style of the hook title shown at the start of each clip; none hides it.
language
string
optional
Two-letter language code of the speech, e.g. en or ru. Omit to auto-detect. Never pass 'auto'.
prompt
string
optional
What kind of moments to pick, in plain words, e.g. 'funny moments about money'.
cold_open
boolean
optional
Start each clip with its strongest moment as a teaser, and also produce standalone hook clips and a looped teaser.
story_arc_style
string
optional
Build each clip as a story (cold open, main part, payoff). Any value except none also turns on cold_open.
add_music
boolean
optional
Add quiet background music from the MnogoReels library.
emojis
boolean
optional
Add emojis to captions where they fit.
smart_cut
boolean
optional
Remove pauses and filler words inside clips. Works only with video_mode podcast (general-beta always does it); ignored for other modes.
b_roll_stock_video
boolean
optional
Insert relevant stock video B-roll.
b_roll_ai_images
boolean
optional
Insert AI-generated illustrative images as B-roll.
b_roll_motion_graphics
boolean
optional
Insert AI motion-graphics B-roll (animated titles, charts, diagrams).
outro
string
optional
Ending added to every clip: default branded outro, soon/like short Russian text cards. Omit to keep the original job's outro (including a custom one).
framing
string
optional
auto = smart crop; full_frame = keep the whole wide frame on a blurred background; portrait_only = always crop to the subject.
agentic_mode
boolean
optional
Extra AI review pass that refines where clips start and end. Slower.
Raw schema
{
"type": "object",
"properties": {
"job_id": {
"type": "string",
"pattern": "^[0-9a-f]{32}$",
"description": "Clip job to recut."
},
"max_credits": {
"type": "integer",
"minimum": 1,
"maximum": 100000,
"description": "Spending limit for this call in credits. Required. Use the credits value from quote_clip_job after the user agreed; the call is refused without charging if the price (or its upper bound) is higher."
},
"start_sec": {
"description": "Start of the part to process, in seconds from the beginning. Default 0.",
"type": "number",
"minimum": 0,
"maximum": 86400
},
"end_sec": {
"description": "End of the part to process, in seconds. Default: the end of the video (its length is looked up automatically when the platform reports it). At most 30 minutes after start_sec.",
"type": "number",
"minimum": 1,
"maximum": 86400
},
"new_moments": {
"default": false,
"description": "true = the AI picks different moments; false = same moments with the new look.",
"type": "boolean"
},
"allow_duplicate": {
"default": false,
"description": "Set true only if the user explicitly wants a second identical job within 10 minutes (it is charged again).",
"type": "boolean"
},
"video_mode": {
"description": "How the video was filmed: general, podcast (talking heads), general-beta (SmartCut: removes pauses and filler words), stream (screen + webcam), vlog, fitness.",
"type": "string",
"enum": [
"general",
"podcast",
"general-beta",
"stream",
"vlog",
"fitness"
]
},
"clip_length": {
"description": "Target length of each clip. auto lets the AI decide; 30_60s suits Shorts, Reels and TikTok.",
"type": "string",
"enum": [
"auto",
"up_to_30s",
"30_60s",
"60_90s",
"90s_3m",
"3m_5m",
"5m_10m"
]
},
"caption_style": {
"description": "Burned-in caption style; empty means no captions. Call get_options_guide to see what each style looks like.",
"type": "string",
"enum": [
"karaoke",
"karaoke_static",
"minimal",
"viral",
"montserrat",
"gilroy",
"rubik",
"bebas",
"impact",
"pixel",
"gothic",
"going",
"hormozi",
"hormozi_pop",
"boxed",
"boxed_black",
"boxed_pop",
"neon",
"pop3d",
"tiktok",
"money",
"chunky",
"punch",
"empty"
]
},
"hook_style": {
"description": "Style of the hook title shown at the start of each clip; none hides it.",
"type": "string",
"enum": [
"default",
"dark",
"accent",
"sticker",
"marker",
"neon",
"stagger",
"torn",
"keyword",
"impact",
"none"
]
},
"language": {
"description": "Two-letter language code of the speech, e.g. en or ru. Omit to auto-detect. Never pass 'auto'.",
"type": "string",
"pattern": "^[a-z]{2}$"
},
"prompt": {
"description": "What kind of moments to pick, in plain words, e.g. 'funny moments about money'.",
"type": "string",
"minLength": 1,
"maxLength": 1000
},
"cold_open": {
"description": "Start each clip with its strongest moment as a teaser, and also produce standalone hook clips and a looped teaser.",
"type": "boolean"
},
"story_arc_style": {
"description": "Build each clip as a story (cold open, main part, payoff). Any value except none also turns on cold_open.",
"type": "string",
"enum": [
"none",
"subtle",
"dynamic",
"cinematic"
]
},
"add_music": {
"description": "Add quiet background music from the MnogoReels library.",
"type": "boolean"
},
"emojis": {
"description": "Add emojis to captions where they fit.",
"type": "boolean"
},
"smart_cut": {
"description": "Remove pauses and filler words inside clips. Works only with video_mode podcast (general-beta always does it); ignored for other modes.",
"type": "boolean"
},
"b_roll_stock_video": {
"description": "Insert relevant stock video B-roll.",
"type": "boolean"
},
"b_roll_ai_images": {
"description": "Insert AI-generated illustrative images as B-roll.",
"type": "boolean"
},
"b_roll_motion_graphics": {
"description": "Insert AI motion-graphics B-roll (animated titles, charts, diagrams).",
"type": "boolean"
},
"outro": {
"description": "Ending added to every clip: default branded outro, soon/like short Russian text cards. Omit to keep the original job's outro (including a custom one).",
"type": "string",
"enum": [
"none",
"default",
"soon",
"like"
]
},
"framing": {
"description": "auto = smart crop; full_frame = keep the whole wide frame on a blurred background; portrait_only = always crop to the subject.",
"type": "string",
"enum": [
"auto",
"full_frame",
"portrait_only"
]
},
"agentic_mode": {
"description": "Extra AI review pass that refines where clips start and end. Slower.",
"type": "boolean"
}
},
"required": [
"job_id",
"max_credits"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
retry_clip_job
Re-runs a failed clip job with the same settings, free of charge, keeping the same job_id. Use it when get_clip_job shows status failed, for example after a temporary download or rendering error.
It refuses jobs that are queued or processing (wait for them) and completed jobs (re-running would replace their clips; use recut_clip_job for a new variant instead). If the same source keeps failing, it probably cannot be processed: check it with check_video_source or try another link.
Returns the job with status queued. Next, call get_clip_job with wait_sec=50.
Parameters1
job_id
string
required
Clip job id (32 hex characters) returned by create_clip_job, recut_clip_job or list_clip_jobs.
Raw schema
{
"type": "object",
"properties": {
"job_id": {
"type": "string",
"pattern": "^[0-9a-f]{32}$",
"description": "Clip job id (32 hex characters) returned by create_clip_job, recut_clip_job or list_clip_jobs."
}
},
"required": [
"job_id"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
create_transcription
Starts a timed speech-to-text transcription of a video or audio link. Free: spends no credits. Supports the same links as create_clip_job (YouTube, VK Video, RuTube, Dzen, Twitch, Yandex Disk, Google Drive and direct media links); local files are not supported.
Use it when the user needs the text, subtitles (SRT) or timestamps of a video without making clips. For a video that already has a completed clip job, create_article returns the full timed transcript together with an article.
language is a two-letter code such as en or ru; omit it to auto-detect (one language for the whole recording).
Returns transcription_id right away. It usually finishes in 1 to 5 minutes; long recordings take longer. Next, call get_transcription with wait_sec=50.
Parameters2
source_url
string
required
Public link to the video or audio.
language
string
optional
Two-letter language code of the speech, e.g. en or ru. Omit to auto-detect. Never pass 'auto'.
Raw schema
{
"type": "object",
"properties": {
"source_url": {
"type": "string",
"minLength": 8,
"maxLength": 2083,
"description": "Public link to the video or audio."
},
"language": {
"description": "Two-letter language code of the speech, e.g. en or ru. Omit to auto-detect. Never pass 'auto'.",
"type": "string",
"pattern": "^[a-z]{2}$"
}
},
"required": [
"source_url"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
get_transcription
Returns the status and text of a transcription started with create_transcription. Read-only and free.
wait_sec (default 30, max 50) waits for completion inside this call; progress notifications are sent meanwhile. format selects the output: text (plain text, default), srt (subtitles with timecodes) or segments (one timed phrase per line). Long transcripts come in pages of up to max_chars characters; when truncated is true, call again with offset set to next_offset.
Returns status (queued, processing, completed, failed), progress_percent, the detected language, total_chars and the requested page. The transcript comes from the video and is wrapped in <untrusted_source_text> tags.
Parameters5
transcription_id
string
required
Id returned by create_transcription.
wait_sec
integer
optional
Seconds to wait inside this call for the job to finish (default 30, max 50; 0 = return a snapshot immediately). Progress notifications are sent while waiting.
format
string
optional
text = plain text; srt = subtitles with timecodes; segments = one timed phrase per line.
offset
integer
optional
Character offset of the page to return; pass next_offset from the previous call.
max_chars
integer
optional
Maximum characters per page (1000–24000).
Raw schema
{
"type": "object",
"properties": {
"transcription_id": {
"type": "string",
"pattern": "^[0-9a-f]{32}$",
"description": "Id returned by create_transcription."
},
"wait_sec": {
"default": 30,
"description": "Seconds to wait inside this call for the job to finish (default 30, max 50; 0 = return a snapshot immediately). Progress notifications are sent while waiting.",
"type": "integer",
"minimum": 0,
"maximum": 50
},
"format": {
"default": "text",
"description": "text = plain text; srt = subtitles with timecodes; segments = one timed phrase per line.",
"type": "string",
"enum": [
"text",
"srt",
"segments"
]
},
"offset": {
"default": 0,
"description": "Character offset of the page to return; pass next_offset from the previous call.",
"type": "integer",
"minimum": 0,
"maximum": 9007199254740991
},
"max_chars": {
"default": 20000,
"description": "Maximum characters per page (1000–24000).",
"type": "integer",
"minimum": 1000,
"maximum": 24000
}
},
"required": [
"transcription_id"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
create_article
Turns the source video of a completed clip job into a long-form article: title, SEO title and description, summary, sections, key takeaways, chapters with timestamps, up to 4 frames from the video as illustrations, and the full timed transcript. Free: spends no credits.
Use it for blog posts, show notes, chapter lists or the full transcript of a video that has already been clipped. It needs a clip job with status completed (create one with create_clip_job first). There is one article per source video per account: calling it again with any clip job of the same video returns the existing article instead of creating a new one; an article that failed cannot be regenerated through the API.
Returns article_id, status and the clip job the article was made from. Writing usually takes 1 to 3 minutes. Next, call get_article with wait_sec=50.
Parameters1
clip_job_id
string
required
Completed clip job whose source video the article is about (32 hex characters).
Raw schema
{
"type": "object",
"properties": {
"clip_job_id": {
"type": "string",
"pattern": "^[0-9a-f]{32}$",
"description": "Completed clip job whose source video the article is about (32 hex characters)."
}
},
"required": [
"clip_job_id"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
get_article
Returns an article created with create_article as Markdown, with its status. Read-only and free.
wait_sec (default 30, max 50) waits inside this call until the text is ready. The Markdown contains the title, summary, sections, key takeaways and chapter timestamps; set include_transcript=true to append the full timed transcript. Long content is paged: when truncated is true, call again with offset set to next_offset.
Returns status (queued, processing, completed, failed), progress_percent, the Markdown page as soon as the text exists (images_pending=true means illustrations are still being prepared; the text will not change), the SEO title and description, image links with captions and source video details. Images that failed cannot be retried through this server. The article is generated from the video and wrapped in <untrusted_source_text> tags.
Parameters5
article_id
string
required
article_id returned by create_article.
wait_sec
integer
optional
Seconds to wait inside this call for the job to finish (default 30, max 50; 0 = return a snapshot immediately). Progress notifications are sent while waiting.
include_transcript
boolean
optional
true appends the full timed transcript (paged).
offset
integer
optional
Character offset of the page to return; pass next_offset from the previous call.
max_chars
integer
optional
Maximum characters per page (1000–24000).
Raw schema
{
"type": "object",
"properties": {
"article_id": {
"type": "string",
"pattern": "^[0-9a-f]{32}$",
"description": "article_id returned by create_article."
},
"wait_sec": {
"default": 30,
"description": "Seconds to wait inside this call for the job to finish (default 30, max 50; 0 = return a snapshot immediately). Progress notifications are sent while waiting.",
"type": "integer",
"minimum": 0,
"maximum": 50
},
"include_transcript": {
"default": false,
"description": "true appends the full timed transcript (paged).",
"type": "boolean"
},
"offset": {
"default": 0,
"description": "Character offset of the page to return; pass next_offset from the previous call.",
"type": "integer",
"minimum": 0,
"maximum": 9007199254740991
},
"max_chars": {
"default": 20000,
"description": "Maximum characters per page (1000–24000).",
"type": "integer",
"minimum": 1000,
"maximum": 24000
}
},
"required": [
"article_id"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
search_broll
Searches B-roll videos and images. catalog=mine searches the user's own library (stock clips, AI images and motion graphics created for their past clip jobs) and returns temporary download links. catalog=shared searches the public MnogoReels stock catalog of about 5,000 reviewed items and returns titles, tags, previews and a website link; shared items are downloaded on the website. Read-only and free.
query matches whole words in titles, descriptions and tags (no stemming); omit it to list the newest items. kind filters image, video or motion. Pages with limit (1 to 50, default 10) and cursor from next_cursor.
Media links expire after 15 minutes; call get_broll_asset for fresh ones. Titles, descriptions, tags and attribution come from videos and stock providers and are wrapped in <untrusted_source_text> tags.
Parameters5
catalog
string
required
mine = your own B-roll; shared = public stock catalog.
query
string
optional
Whole words matched in titles, descriptions and tags (no stemming); omit to list the newest items.
kind
string
optional
image, video or motion (motion graphics); omit for all.
limit
integer
optional
Items per page, 1–50 (default 10).
cursor
string
optional
next_cursor from the previous page.
Raw schema
{
"type": "object",
"properties": {
"catalog": {
"type": "string",
"enum": [
"mine",
"shared"
],
"description": "mine = your own B-roll; shared = public stock catalog."
},
"query": {
"description": "Whole words matched in titles, descriptions and tags (no stemming); omit to list the newest items.",
"type": "string",
"minLength": 1,
"maxLength": 500
},
"kind": {
"description": "image, video or motion (motion graphics); omit for all.",
"type": "string",
"enum": [
"image",
"video",
"motion"
]
},
"limit": {
"default": 10,
"description": "Items per page, 1–50 (default 10).",
"type": "integer",
"minimum": 1,
"maximum": 50
},
"cursor": {
"description": "next_cursor from the previous page.",
"type": "string",
"minLength": 1,
"maxLength": 512
}
},
"required": [
"catalog"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
get_broll_asset
Returns one B-roll item by asset_id with fresh links. Read-only and free; needs an API key. Use it right before downloading an item found with search_broll, because catalog=mine download links expire after 15 minutes; to find items use search_broll instead. catalog must be the catalog the asset_id came from.
For catalog=mine it returns media_url and poster_url, valid for links_expire_in_sec seconds. For catalog=shared it returns title, tags, a preview poster (videos only), attribution and web_url, because shared items are downloaded on the MnogoReels website. Text from videos and stock providers is wrapped in <untrusted_source_text> tags.
Parameters2
catalog
string
required
The catalog the asset_id came from in search_broll: mine or shared.
asset_id
string
required
asset_id from search_broll (64 hex characters).
Raw schema
{
"type": "object",
"properties": {
"catalog": {
"type": "string",
"enum": [
"mine",
"shared"
],
"description": "The catalog the asset_id came from in search_broll: mine or shared."
},
"asset_id": {
"type": "string",
"pattern": "^[0-9a-f]{64}$",
"description": "asset_id from search_broll (64 hex characters)."
}
},
"required": [
"catalog",
"asset_id"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
list_packs
Lists the user's publication packs, newest first. A pack is made on the MnogoReels website from one video link and bundles clips, a transcript, suggested titles, descriptions and hashtags, YouTube thumbnails, an article and teasers. Read-only and free; packs cannot be created through the API.
Pages with limit (1 to 50, default 10) and offset. Returns pack_id, status, progress, source_url, created_at, the status of each part and clip_job_id (use it with get_clip_job). Next, call get_pack for titles, hashtags, thumbnails and teasers.
Parameters2
limit
integer
optional
Items per page, 1–50 (default 10).
offset
integer
optional
Items to skip; pass next_offset from the previous page.
Returns one publication pack: suggested titles, descriptions, cover texts and hashtags, thumbnail images, teaser videos, the article id and the clip job id. Read-only and free.
Set include_transcript=true to add the transcript, paged with offset and next_offset. Use get_clip_job with clip_job_id for the clips and get_article with article_id for the article. Text generated from the video is wrapped in <untrusted_source_text> tags.
Parameters4
pack_id
string
required
pack_id from list_packs (32 hex characters).
include_transcript
boolean
optional
true adds the transcript (paged with offset and max_chars).
offset
integer
optional
Character offset of the page to return; pass next_offset from the previous call.
max_chars
integer
optional
Maximum characters per page (1000–20000).
Raw schema
{
"type": "object",
"properties": {
"pack_id": {
"type": "string",
"pattern": "^[0-9a-f]{32}$",
"description": "pack_id from list_packs (32 hex characters)."
},
"include_transcript": {
"default": false,
"description": "true adds the transcript (paged with offset and max_chars).",
"type": "boolean"
},
"offset": {
"default": 0,
"description": "Character offset of the page to return; pass next_offset from the previous call.",
"type": "integer",
"minimum": 0,
"maximum": 9007199254740991
},
"max_chars": {
"default": 12000,
"description": "Maximum characters per page (1000–20000).",
"type": "integer",
"minimum": 1000,
"maximum": 20000
}
},
"required": [
"pack_id"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
create_youtube_thumbnails
Generates 1 to 4 AI YouTube thumbnail images (1280×720 JPEG) for a video, using frames from the video plus an optional title and art direction. Free: spends no credits, but every variant is a separate AI image generation, so ask for more than one only when the user wants options.
Parameters: source_url (link to the video), language of the text on the thumbnail (required; match the title or the video), title (text to place on the thumbnail, up to 120 characters), prompt (art direction, up to 1000 characters), style (auto, bold, cinematic, minimal, news, gaming) and variants (1 to 4, default 1).
Returns thumbnail_job_id right away. It usually finishes in 1 to 2 minutes. Next, call get_youtube_thumbnails with wait_sec=50.
Parameters6
source_url
string
required
Public link to the video: YouTube, VK Video, RuTube, Dzen, Twitch, Yandex Disk file, Google Drive file, or a direct .mp4/.mov/.mkv/.webm link. Local files are not supported.
language
string
required
Two-letter language of the text on the thumbnail, e.g. en; match the title or the video.
title
string
optional
Text printed on the thumbnail, ≤120 characters; omit to let the AI write it.
prompt
string
optional
Art direction, e.g. "surprised face, red arrow, dark background".
style
string
optional
auto, bold, cinematic, minimal, news or gaming.
variants
integer
optional
Number of images, 1–4 (each is a separate AI generation).
Raw schema
{
"type": "object",
"properties": {
"source_url": {
"type": "string",
"minLength": 8,
"maxLength": 2083,
"description": "Public link to the video: YouTube, VK Video, RuTube, Dzen, Twitch, Yandex Disk file, Google Drive file, or a direct .mp4/.mov/.mkv/.webm link. Local files are not supported."
},
"language": {
"type": "string",
"pattern": "^[a-z]{2}$",
"description": "Two-letter language of the text on the thumbnail, e.g. en; match the title or the video."
},
"title": {
"description": "Text printed on the thumbnail, ≤120 characters; omit to let the AI write it.",
"type": "string",
"minLength": 1,
"maxLength": 120
},
"prompt": {
"description": "Art direction, e.g. \"surprised face, red arrow, dark background\".",
"type": "string",
"minLength": 1,
"maxLength": 1000
},
"style": {
"default": "auto",
"description": "auto, bold, cinematic, minimal, news or gaming.",
"type": "string",
"enum": [
"auto",
"bold",
"cinematic",
"minimal",
"news",
"gaming"
]
},
"variants": {
"default": 1,
"description": "Number of images, 1–4 (each is a separate AI generation).",
"type": "integer",
"minimum": 1,
"maximum": 4
}
},
"required": [
"source_url",
"language"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
get_youtube_thumbnails
Returns the status and images of a thumbnail job started with create_youtube_thumbnails. Read-only and free; needs an API key. Jobs cannot be listed, so keep the thumbnail_job_id from create_youtube_thumbnails.
wait_sec (default 30, max 50) waits inside this call until the job completes or fails, with progress notifications; wait_sec=0 returns a snapshot. Returns status (queued, processing, completed, failed), stage, progress_percent and the finished variants (1280×720 JPEG image link, width, height). If some variants failed, the rest are still returned.
Parameters2
thumbnail_job_id
string
required
thumbnail_job_id returned by create_youtube_thumbnails.
wait_sec
integer
optional
Seconds to wait inside this call for the job to finish (default 30, max 50; 0 = return a snapshot immediately). Progress notifications are sent while waiting.
Raw schema
{
"type": "object",
"properties": {
"thumbnail_job_id": {
"type": "string",
"pattern": "^[0-9a-f]{32}$",
"description": "thumbnail_job_id returned by create_youtube_thumbnails."
},
"wait_sec": {
"default": 30,
"description": "Seconds to wait inside this call for the job to finish (default 30, max 50; 0 = return a snapshot immediately). Progress notifications are sent while waiting.",
"type": "integer",
"minimum": 0,
"maximum": 50
}
},
"required": [
"thumbnail_job_id"
],
"$schema": "https://json-schema.org/draft/2020-12/schema"
}
MnogoReels MCP server: turn long videos into captioned vertical Shorts/Reels, transcripts and articles from any MCP client. Give your AI assistant a YouTube, VK Video, RuTube, Dzen, Twitch, Yandex Disk, Google Drive or direct video link, and it picks the best moments, reframes them to 9:16, burns in captions and a hook title, and hands back links to the finished clips, using your own MnogoReels account.
Cursor and Kiro buttons expect MNOGOREELS_API_KEY in your environment; set it and restart the editor, or replace ${env:…} with your key in the config the editor opens. VS Code buttons ask for the key.
Direct app links (if a button does not open your editor)
GitHub does not render cursor:// and vscode: links as clickable, so copy one into your browser's address bar.
The Claude Desktop bundle (.mcpb) opens an install dialog that asks for the key and an optional read-only switch. Goose links and every other client are listed under Quick start.
Put it into your client configuration as shown below (MNOGOREELS_API_KEY for the local server, an Authorization: Bearer header for the hosted one).
The key can spend your credits. Use a separate key for each client or machine, never paste it into chats, issues or URLs, and revoke it on the same page when you no longer need it.
Listing tools works without a key, and so do get_account, get_options_guide, check_video_source and quote_clip_job for a new link. Everything that reads your jobs or creates new ones needs the key.
Quick start
Hosted server (recommended)
URL: https://mcp.mnogo-reels.ru/mcp (Streamable HTTP). Send the key in a header:
"type": "http" is the spelling used by Claude Code, VS Code and Cursor. Cline needs "streamableHttp", Windsurf serverUrl, Gemini CLI httpUrl; the per-client blocks below have the exact form. Ready-made files are in examples/, and docs/clients.md has longer notes per client.
Download mnogoreels-mcp.mcpb and double-click it (or drag it into Settings → Extensions). Claude Desktop asks for the API key and runs the server with its built-in Node.js.
Or add the local server to claude_desktop_config.json (Settings → Developer → Edit Config):
Local variant: the local server block above, with "MNOGOREELS_API_KEY": "${env:MNOGOREELS_API_KEY}" if you keep the key in your environment.
VS Code (GitHub Copilot)
Use the buttons above, or create .vscode/mcp.json. VS Code asks for the key once and stores it securely:
json
{"inputs":[{"type":"promptString","id":"mnogoreels_api_key","description":"MnogoReels API key (mrfrt_usr_...)","password":true}],"servers":{"mnogoreels":{"type":"http","url":"https://mcp.mnogo-reels.ru/mcp","headers":{"Authorization":"Bearer ${input:mnogoreels_api_key}"}}}}
A workspace .mcp.json uses the mcpServers form from Hosted server; do not commit a file that contains your key.
OpenAI Codex
~/.codex/config.toml, hosted (export MNOGOREELS_API_KEY in the shell that starts Codex):
Never omit type for the remote entry: Cline treats a remote entry without type as legacy SSE, which this server does not serve. Cline can also set the server up by itself from llms-install.md.
Goose
Paste one of these links into your browser's address bar (GitHub does not render goose:// links as clickable). Local, Goose asks for the key:
Or run goose configure → Add Extension → Command-line Extension with the command npx -y mnogoreels-mcp@latest and the variable MNOGOREELS_API_KEY. A config.yaml example is in examples/goose-config.yaml.
Kiro
Use the buttons above, or add to ~/.kiro/settings/mcp.json (or .kiro/settings/mcp.json in a project):
These clients accept the generic mcpServers JSON: paste the hosted or the local block into their MCP settings (JetBrains: Settings → Tools → AI Assistant → Model Context Protocol; Warp: Settings → AI → MCP servers; Amazon Q: ~/.aws/amazonq/mcp.json; LM Studio: mcp.json). If a client does not support remote servers with headers, use the local block.
Continue
Save as .continue/mcpServers/mnogoreels.yaml and add the secret MNOGOREELS_API_KEY in Continue:
"How many credits would it cost to clip the first 30 minutes of this podcast? It's 2 hours long."
"Transcribe this RuTube video and give me SRT subtitles."
"Write a blog post with chapters from my last clip job."
"Show my recent clip jobs and the three best clips of each, with their scores and links."
Tools
Tool
Title
Read-only
Spends credits
What it does
get_account
Check API key and settings
yes
no
Checks whether the configured MnogoReels API key works and shows this server's settings.
get_options_guide
Explain clip options
yes
no
Explains the option values accepted by create_clip_job and recut_clip_job: video modes, clip lengths, caption styles, hook styles, story arc styles, outros, framing and languages, plus the pricing rules and the 30-minute limit per job.
check_video_source
Check video source
yes
no
Checks whether MnogoReels can download a video link and returns its title, length and platform.
quote_clip_job
Quote clip job cost
yes
no
Calculates the credit cost of create_clip_job or recut_clip_job without spending anything, and checks that MnogoReels can download the video.
create_clip_job
Create clip job (spends credits)
no
yes
Starts AI clipping of a long video into vertical 9:16 short clips with burned-in captions and a hook title, plus optional cold-open hooks, music and B-roll.
get_clip_job
Get clip job status and clips
yes
no
Returns the status, progress and results of a clip job created by create_clip_job, recut_clip_job or on the MnogoReels website.
list_clip_jobs
List clip jobs
yes
no
Lists the user's clip jobs, newest first, including jobs started on the MnogoReels website.
recut_clip_job
Recut clip job (spends credits)
no
yes
Creates a new clip job from the same source video as an existing job, with different settings such as caption style, hook style, clip length, language, framing, music, B-roll or prompt.
retry_clip_job
Retry failed clip job (free)
no
no
Re-runs a failed clip job with the same settings, free of charge, keeping the same job_id.
create_transcription
Transcribe video or audio
no
no
Starts a timed speech-to-text transcription of a video or audio link.
get_transcription
Get transcription
yes
no
Returns the status and text of a transcription started with create_transcription.
create_article
Create article from clip job
no
no
Turns the source video of a completed clip job into a long-form article: title, SEO title and description, summary, sections, key takeaways, chapters with timestamps, up to 4 frames from the video as illustrations, and the full timed transcript.
get_article
Get article
yes
no
Returns an article created with create_article as Markdown, with its status.
search_broll
Search B-roll library
yes
no
Searches B-roll videos and images.
get_broll_asset
Get B-roll asset
yes
no
Returns one B-roll item by asset_id with fresh links.
list_packs
List publication packs
yes
no
Lists the user's publication packs, newest first.
get_pack
Get publication pack
yes
no
Returns one publication pack: suggested titles, descriptions, cover texts and hashtags, thumbnail images, teaser videos, the article id and the clip job id.
create_youtube_thumbnails
Create YouTube thumbnails
no
no
Generates 1 to 4 AI YouTube thumbnail images (1280×720 JPEG) for a video, using frames from the video plus an optional title and art direction.
get_youtube_thumbnails
Get YouTube thumbnails
yes
no
Returns the status and images of a thumbnail job started with create_youtube_thumbnails.
Every tool declares a title, all four MCP annotations (readOnlyHint, destructiveHint, idempotentHint, openWorldHint) and an output schema. The two credit-spending tools are marked destructiveHint: true, so clients that ask before risky actions (for example Claude and Codex) prompt you on every call.
How to choose a tool
I want to…
Tool
Know whether a link works and how long the video is
check_video_source
Know the price before making clips
quote_clip_job
Make clips
create_clip_job (after the quote and your OK)
Follow progress and get the clips
get_clip_job
Find my earlier jobs (also those made on the website)
list_clip_jobs
Get a different look or different moments from a finished job
recut_clip_job (paid, new job)
Re-run a failed job
retry_clip_job (free)
Get a transcript or SRT subtitles
create_transcription → get_transcription
Get a blog post, show notes or chapters
create_article → get_article (needs a completed clip job)
Understand caption styles, modes and other options
get_options_guide
Jobs run in the background
Every create_* tool returns an id right away. The matching get_* tool waits inside the call for up to wait_sec seconds (default 30, max 50) and sends progress notifications meanwhile, so the assistant simply calls it again until the status is completed or failed. Short videos take a few minutes; a full 30-minute range can take tens of minutes. While a clip job is processing, get_clip_job returns only the number of ready clips (set include_partial_clips=true to see them early).
Spending rules
Price: 1 credit per started minute of the processed range (end_sec − start_sec), minimum 1, charged when the job is created. Examples: 45 s → 1 credit, 10 min → 10, 29 min 30 s → 30. Captions, hooks, B-roll, music and agentic mode do not change the price through the API.
Free: checks, quotes, transcriptions, articles, YouTube thumbnails, retries of failed jobs and reading results.
Quote first:quote_clip_job is free and uses the same formula as the backend. The assistant should show you the price and wait for your agreement.
max_credits is required on create_clip_job and recut_clip_job. The call is refused without charging if the price is higher. An operator can also set a hard cap with MNOGOREELS_MAX_CREDITS_PER_CALL or the X-MnogoReels-Max-Credits-Per-Call header that the model cannot raise.
30 minutes per job. MnogoReels processes at most 30 minutes per job, so longer ranges are refused before any charge. Split long videos with start_sec/end_sec (0–1800, 1800–3600, …).
Duplicate guard: an identical request with the same key within 10 minutes returns the job that was already created instead of charging again (allow_duplicate: true turns this off). The guard lives in one server process; two separate processes do not share it.
No cancel, no automatic refund. Credits are charged when the job is created and jobs cannot be cancelled. If processing fails, credits are not returned automatically; retry_clip_job re-runs a failed job for free, or contact support@mnogo-reels.ru with the job_id.
Confirmation dialog. Where the client supports MCP form elicitation, the server also asks you to confirm the exact amount before charging:
Where the server runs
Confirmation dialog before spending
Local server (stdio, npx, .mcpb, Docker), any client with form elicitation
yes
Hosted server, client on MCP 2026-07-28 with form elicitation
no: max_credits, the operator cap, the 30-minute limit, the duplicate guard and your client's own approval prompt are the gates
Any client without elicitation support
no: same gates as above
Prompts and resources
Prompts (slash commands in clients that show them):
Prompt
What it does
clip_video
Quote, confirm, create and wait for clips from one link (url, optional language, caption_style, clip_length)
podcast_to_shorts
Split a long podcast into 30-minute parts, quote each, make a Shorts series and propose a posting order (url, optional language)
video_to_blog_post
Turn a completed clip job into an article plus a chapter list (clip_job_id)
transcribe_video
Transcribe a link as text or SRT (url, optional language, format)
review_recent_clips
Review your latest jobs and suggest recut settings without spending anything (optional count)
Resources:
URI
Content
mnogoreels://guide/options
All option values with descriptions
mnogoreels://guide/pricing
Pricing, the 30-minute limit and the list of free operations
mnogoreels://guide/workflow
How to use the tools (same text as the server instructions)
mnogoreels://clip-jobs/{job_id}
Status and clips of one job (JSON; lists your 20 latest jobs)
mnogoreels://transcriptions/{transcription_id}
First page of a transcript
mnogoreels://articles/{article_id}
First page of an article in Markdown
Prompts and job resources appear only when the tools they use are enabled (for example, /mcp/readonly offers only review_recent_clips). Every capability is also reachable through tools, because many clients never show resources.
Configuration
Environment variables (local server and self-hosting)
Variable
Default
Meaning
MNOGOREELS_API_KEY
none
Your key (mrfrt_usr_…). Not needed to start the server or list tools. Ignored in HTTP mode, where every caller sends their own key.
Comma list of clips, transcripts, articles, library, thumbnails (or all). account is always on.
MNOGOREELS_READ_ONLY
false
true hides every tool that creates jobs or spends credits.
MNOGOREELS_MAX_CREDITS_PER_CALL
unset
Hard cap on credits per paid call; the model cannot override it.
MNOGOREELS_MAX_WINDOW_SEC
1800
Longest range per paid job, in seconds. Keep the default: it matches what MnogoReels processes.
MNOGOREELS_MEDIA_URLS
direct
both also returns stream_url links through the MnogoReels website proxy. Keep direct until MnogoReels announces the proxy for API use.
MNOGOREELS_DEBUG
false
Local server only: log full error messages to stderr.
MNOGOREELS_TRANSPORT
stdio
http starts the Streamable HTTP server (same as --http).
MNOGOREELS_HOST / MNOGOREELS_PORT
127.0.0.1 / 8080
HTTP bind address.
MNOGOREELS_PUBLIC_URL
https://mcp.mnogo-reels.ru/mcp
Public URL announced in the server card.
MNOGOREELS_ALLOWED_HOSTS
localhost,127.0.0.1,[::1]
Accepted Host headers (DNS-rebinding protection).
MNOGOREELS_ALLOWED_ORIGINS
localhost,127.0.0.1,[::1]
Accepted browser Origin hostnames, or * for a public bearer-token deployment.
MNOGOREELS_TRUSTED_PROXIES
empty
CIDRs of reverse proxies whose X-Forwarded-For is trusted.
Booleans accept 1/true/yes/on and 0/false/no/off. An invalid value stops the server with a message naming the variable (exit code 2). Command line: mnogoreels-mcp [--http] [--help] [--version].
Hosted server: paths and headers
Path
Profile
/mcp
default
/mcp/readonly
readonly
/mcp/directory
directory
Header
Effect
Authorization: Bearer <key>
Your API key (preferred).
X-API-Key: <key>
Your API key, when Authorization is not available.
X-MnogoReels-Toolsets: clips,transcripts
Show only these toolsets (account is always kept).
X-MnogoReels-Read-Only: true
Hide tools that create jobs or spend credits. false never widens a read-only path.
X-MnogoReels-Max-Credits-Per-Call: 30
Hard credit cap per paid call.
Headers can only narrow what the path allows. Discovery and listing work without a key; a tool call that needs the key and has none gets HTTP 401.
Profiles
Profile
Tools
What changes
default
19
Everything.
readonly
13
No tool that creates jobs or spends credits; reading, checks and quotes still work.
directory
17
No YouTube thumbnail tools, no AI-image or motion-graphics B-roll options, no "buy credits" link. Used for public directory listings.
Self-hosting the HTTP server
npx -y mnogoreels-mcp@latest --http or docker run -p 8080:8080 -e MNOGOREELS_TRANSPORT=http ghcr.io/realsatanyan/mnogoreels-mcp:latest starts the same server that runs at mcp.mnogo-reels.ru (health check: GET /healthz, server card: GET /.well-known/mcp/server-card.json). Set MNOGOREELS_ALLOWED_HOSTS to your domain. The production setup is described in docs/deploy-coolify.md.
Security
Untrusted content. Titles, transcripts, articles and other text that comes from videos is wrapped in <untrusted_source_text> tags, and the server instructions tell the model to treat it as data, never as instructions.
Key hygiene. The key travels only in the Authorization or X-API-Key header (the hosted server rejects keys in URLs), is never logged, and get_account shows only its last 4 characters. Use a dedicated key per client and revoke keys you no longer use. Service keys (mrfrt_svc_…) are rejected.
Read-only mode.MNOGOREELS_READ_ONLY=true, the /mcp/readonly path or the read-only switch in the Claude Desktop bundle removes every tool that can create jobs or spend credits.
Only official sources. The server is published only as the npm package mnogoreels-mcp, the image ghcr.io/realsatanyan/mnogoreels-mcp, the hosted endpoint https://mcp.mnogo-reels.ru and this repository. Do not give your key to look-alikes.
Pinning.mnogoreels-mcp@latest updates automatically. To avoid automatic updates, pin a version, e.g. npx -y mnogoreels-mcp@1.0.0.
Report vulnerabilities privately: see SECURITY.md.
Privacy
The local server sends your tool arguments and key only to api.mnogo-reels.ru. The hosted server receives them at mcp.mnogo-reels.ru and forwards them to the same API; it keeps keys only in memory for the duration of a request and never logs keys, tool arguments or results (only a short hash prefix of the key, to tell users apart). No telemetry is collected. Details: PRIVACY.md, the MnogoReels privacy policy and terms.
Troubleshooting
Tool errors start with Error [<code>]: followed by advice. The most common ones:
Code
What to do
missing_api_key, invalid_api_key_format, invalid_api_key (HTTP 401 on the hosted server)
Create a key at Developers → API keys and update the client config. Keys start with mrfrt_usr_. Run get_account to check.
The platform does not report the video length (typical for direct file links). Pass end_sec (the length in seconds, or the end of the part to process).
invalid_arguments mentioning 30 minutes
One job processes at most 30 minutes. Split the video with start_sec/end_sec and make several jobs.
source_not_downloadable
The video is private, removed, region-locked, audio-only or larger than 17 GB. Nothing was charged. Try another link.
rate_limited, rate_limited_local
Too many requests; wait the number of seconds in the message.
upstream_timeout
MnogoReels did not answer in time; try again. For source checks you can pass start_sec/end_sec yourself.
outcome_unknown, billing_uncertain
The server could not confirm whether a paid job was created. Check list_clip_jobs before trying again; if credits were taken without a job, contact support@mnogo-reels.ru. An identical retry within 10 minutes returns the same answer without sending anything.
Codex: tools time out or the server does not start
Set startup_timeout_sec = 60 and tool_timeout_sec = 120 in config.toml.
Slow or failing downloads in Russia
Some Russian home connections throttle the API and media hosts. Prefer the hosted server on home connections. MNOGOREELS_MEDIA_URLS=both adds proxied stream_url links once MnogoReels enables that proxy for API use.
No local files: the video must be reachable by a public link (upload it somewhere public first).
No credit balance: the API does not expose it; see it on mnogo-reels.ru.
No publishing or scheduling to social networks.
No cancelling of jobs.
Publication packs can be read but not created; failed article images cannot be retried.
Development
Requirements: Node.js 22+. The code is TypeScript (strict), built with tsc, tested with Vitest and Stryker, and written test-first.
bash
npm ci --ignore-scripts
npm run typecheck && npm run lint && npm test
npm run build
npm run test:integration
npm run smoke # MCP Inspector CLI smoke test
npm run version:check # manifests in sync with package.json
npm run metadata:check # README tool tables, MCPB tools, Docker tools.json up to date