youtube-transcripts MCP server
Commit Nov 13, 2026Extract timestamped transcripts and metadata from YouTube videos and Shorts for summarization and searching.
Also known as johnisanerd/alphaosint-mcp-servers
- Stars
- 0
- Tools
- 5
- Downloads, 30 days
- —
- Latest version
- 1.1.1
- Last commit
- Nov 13, 2026
- License
- MIT
What is youtube-transcripts MCP?
youtube-transcripts extracts timestamped transcripts and metadata from YouTube videos and Shorts for summarization and searching. It exposes 5 tools, such as get-actor-run, get-dataset-items, get-key-value-store-record.
Summary generated from the project's README
About youtube-transcripts MCP server
Extracting timestamped text and metadata from YouTube videos or Shorts is what this server enables for developers using AI agents. It provides the actual content of a video rather than just its statistics, making it ideal for tasks involving summarization, searching, or quoting specific moments. The server runs on a hosted Apify endpoint, meaning there is no local installation required. Users connect via a streamable HTTP transport using an Apify OAuth login or a bearer token.
Through the johnvc--YoutubeTranscripts tool, users can process single URLs or arrays of up to 100 links. The server supports language selection, translation to specific ISO 639-1 codes, and filtering for manual or auto-generated captions. It can also retrieve extended metadata such as like counts, upload dates, and categories. Developers can choose various output formats including plain text, SRT, or VTT files. The service is billed through an Apify account based on the number of videos processed and the level of metadata requested.
What you can do with it
- 1Summarize video contentProvide an AI agent with the full transcript of a video to generate a concise summary.
- 2Search for specific quotesFind exact moments in a video by searching through timestamped transcript snippets.
- 3Batch process channel contentAutomatically extract transcripts for the most recent videos from a specific YouTube channel.
Written from the project's README. Every use case is checked against a quote from it.
How to install youtube-transcripts MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall youtube-transcripts MCP in Claude Code
$ claude mcp add --transport http youtube-transcripts 'https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts'Install youtube-transcripts MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"youtube-transcripts": {
"url": "https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts"
}
}
}Install youtube-transcripts MCP in VS Code
.vscode/mcp.json
{
"servers": {
"youtube-transcripts": {
"type": "http",
"url": "https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts"
}
}
}Install youtube-transcripts MCP in VS Code CLI
$ code --add-mcp '{"name":"youtube-transcripts","type":"http","url":"https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts"}'Install youtube-transcripts MCP in Codex
$ codex mcp add youtube-transcripts --url 'https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts'Install youtube-transcripts MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.youtube-transcripts] url = "https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts"
Install youtube-transcripts MCP in Gemini CLI
$ gemini mcp add -s user --transport http youtube-transcripts 'https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts'Install youtube-transcripts MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"youtube-transcripts": {
"url": "https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts",
"type": "http"
}
}
}youtube-transcripts MCP tools 5
Read from the server with tools/listget-actor-runRead-onlyGet detailed information about a specific Actor run. Returns run result: status, storages (datasets/keyValueStores alias map), stats, summary, nextStep. - summary describes the past (e.g. "SUCCEEDED in 22s. 47 items; 3 fields available."). - nextStep prescribes one primary follow-up action with identifiers interpolated (e.g. "Use get-dataset-items with datasetId=..."). - waitSecs (0–45, default 30) waits up to that many seconds for terminal status before returning. USAGE: - Use to check the status of a run started by any Actor-running tool. - Pass waitSecs > 0 to block until terminal (or until the cap elapses). USAGE EXAMPLES: - user_input: Show details of run y2h7sK3Wc - user_input: Wait for run y2h7sK3Wc to finish
get-dataset-itemsRead-onlyGet items (rows) from a dataset — the output/results produced by an Actor run. Returns the rows themselves, not dataset metadata, counts, or a schema. When the user provides a datasetId and asks to retrieve results, output, data, or rows, call this tool directly. Default limit is 20. Use clean=true to skip empty items and hidden fields. USAGE: - Use when you need to read data from a dataset (all items or only selected fields). USAGE EXAMPLES: - user_input: Retrieve results from dataset abc123 - user_input: Get only metadata.url and title from dataset username~my-dataset
get-key-value-store-recordRead-onlyGet the value stored under a specific key in a key-value store — a single record, not a listing of all keys. Requires the exact key name. The response preserves the original Content-Encoding; most clients handle decompression automatically. A record over 262144 bytes is returned as a link to download it, not as its value. USAGE: - Use when you need to retrieve a specific record (JSON, text, or binary) from a store. USAGE EXAMPLES: - user_input: Get record INPUT from store abc123 - user_input: Get record data.json from store username~my-store
abort-actor-runActionAbort an Actor run that is currently starting or running. For runs with status SUCCEEDED, FAILED, ABORTING, ABORTED, or TIMED-OUT, this call has no effect. The results will include the updated run details after the abort request. USAGE: - Use when you need to stop a run that is taking too long or misconfigured. USAGE EXAMPLES: - user_input: Abort run y2h7sK3Wc - user_input: Gracefully abort run y2h7sK3Wc
johnvc--YoutubeTranscriptsActionThis tool calls the Actor "johnvc/YoutubeTranscripts" and retrieves its output results. Actor description: Scrape YouTube transcripts, subtitles, and captions in bulk, the cheapest pay-per-video YouTube transcript API on Apify. Callable from any MCP client (Claude, Cursor, ChatGPT). Supports YouTube videos, Shorts, and every URL format.
Automated checks
Ran Nov 16, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachablejohnisanerd/alphaosint-mcp-servers is public and reachablegithub.com/johnisanerd/alphaosint-mcp-servers ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-11-13github.com/johnisanerd/alphaosint-mcp-servers ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model5 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNo tool named like delete, drop or execPass
- Remote endpoint answers the MCP handshakeinitialize answered with protocol 2025-11-25 in 105 ms; tools/list returned 5 tools (HostedMCP test account)Pass
README
View on GitHubYouTube Transcripts MCP server
Registry name: io.github.johnisanerd/youtube-transcripts. Maintained by AlphaOSINT (Apify developer johnvc, GitHub johnisanerd).
What it does
Get the transcript of one or many YouTube videos or Shorts, with timestamps and language selection, plus optional video metadata.
When to use it
Video text for summarizing, searching and quoting videos. Use it to give an agent the content of a video rather than its statistics.
How it works
The server runs on Apify's hosted MCP endpoint, so there is nothing to install or host. Its URL pins one Actor with ?tools=johnvc/YoutubeTranscripts, which gives your agent a single tool, johnvc--YoutubeTranscripts, whose parameters are the Actor's typed input schema. The agent can call that tool directly without searching for an Actor first, and it can't start other Actors through this server. For general access to the whole Apify Store, use https://mcp.apify.com without the tools parameter.
Pricing
Each tool call runs the Apify Actor johnvc/YoutubeTranscripts on your own Apify account and is billed there at the Actor's prices:
| Charge | Price (USD) | What it covers |
|---|---|---|
| Video Processed | $0.00275 | Charged per YouTube video successfully transcribed. Includes the timestamped and plain-text transcript plus basic video metadata (title, channel, views, duration, thumbnail). Failed videos, videos without captions, and… |
| Extended video metadata | $0.003 | Charged only when include_extended_metadata is true and the extra lookup succeeds (adds like count, upload date, categories and availability). Basic metadata is included free with every transcript. Failures are not char… |
Free-plan rates as of 2026-11-12; paid Apify plans pay less per event. Apify also bills a token amount per run start and stored row ($0.00001, $0.00001). Current prices are on the Actor's page: https://apify.com/johnvc/YoutubeTranscripts
Endpoint
- Transport: Streamable HTTP
- URL:
https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts - Auth: Apify OAuth (opens on first use), or the header
Authorization: Bearer <APIFY_TOKEN>(Apify Console → Settings → API & Integrations)
Setup
Claude Code
claude mcp add --transport http youtube-transcripts "https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts"
VS Code (.vscode/mcp.json)
{
"servers": {
"youtube-transcripts": {
"type": "http",
"url": "https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts"
}
}
}
Cursor (~/.cursor/mcp.json)
{
"mcpServers": {
"youtube-transcripts": {
"url": "https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts"
}
}
}
Claude Desktop: Settings → Connectors → Add custom connector, and paste the URL. Older versions without remote connectors can bridge through mcp-remote in claude_desktop_config.json:
{
"mcpServers": {
"youtube-transcripts": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts"
]
}
}
}
Tools
| Tool | What it does |
|---|---|
johnvc--YoutubeTranscripts |
Runs the Actor with the parameters below and returns the result rows. |
get-actor-run |
Status of a run the tool started (for long runs). |
get-dataset-items |
Pages through a run's full output when it is larger than the inline result. |
get-key-value-store-record |
Reads a record a run stored. |
abort-actor-run |
Stops a run. |
Apify's endpoint adds the four helper tools whenever an Actor tool is loaded.
Parameters of johnvc--YoutubeTranscripts
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
youtube_url |
string or array | "https://www.youtube.com/watch?v=1UL8iZJD38U" |
Required unless 'channel' is provided. Provide one YouTube URL as a string, or multiple as an array (max 100 URLs per run). Works with standard videos, Shorts, youtu.be short links, embed URLs, and m.youtube.com mobile… | |
languages |
array | ["en"] |
Ordered list of ISO 639-1 language codes. The first available transcript matching one of these languages is returned. Example: ["en", "es", "fr"] tries English first, then Spanish, then French. Defaults to ["en"] when o… | |
translate_to |
string | Optional. If set to an ISO 639-1 code (e.g. "es", "fr", "de"), the picked transcript is translated to this language. Requires the source transcript to be translatable - run with list_only=true first to see translation_l… | ||
transcript_type |
string: any, manual, generated | "any" |
Filter which transcript variant to consider when both manually-created and auto-generated captions exist for a language. "any" picks whichever matches the language priority first; "manual" only considers human-made capt… | |
output_formats |
array | [] |
Optional list of extra output formats to include in the dataset alongside the default structured fields. Allowed values: "srt" (SubRip subtitles), "vtt" (WebVTT subtitles), "text" (plain text, newline-separated). Unknow… | |
preserve_formatting |
boolean | false |
If true, keep YouTube's inline HTML formatting tags (, ) in transcript text. Default false (strips them for clean text). | |
list_only |
boolean | false |
If true, do NOT fetch any transcript. Instead, for each URL return the list of all available transcripts with their language code, generated/manual status, is_translatable flag, and translation_languages. Useful for dis… | |
include_metadata |
boolean | true |
If true, attach video metadata from the same YouTube response that carries the captions: title, description, channel_name, channel_id, channel_url, view_count, video_duration_seconds, thumbnail_url, tags, was_live. No e… | |
include_extended_metadata |
boolean | false |
If true, run a second lookup per video for the fields the caption response does not carry: like_count, upload_date, categories, availability. Adds about 1-3 seconds per video and is charged as a videometadata event when… | |
channel |
string or array | Optional. One YouTube channel as a string, or multiple as an array. Accepts an @handle ("@mkbhd"), a bare handle ("mkbhd"), a channel ID ("UCBJycsmduvYEL83R_U4JriQ"), or any channel URL (/channel/, /c/, /user/, /@handle… | ||
max_videos |
integer | 100 |
Maximum number of videos to list per channel, newest first. Default 100. Values above 1000 are capped at 1000 (and values below 1 are raised to 1) with a log warning rather than failing the run. When channel_transcripts… | |
channel_transcripts |
boolean | false |
If true, every video listed from the channel(s) is also queued for transcript extraction (newest first, deduplicated against youtube_url entries, subject to the 100 videos per run limit). Each successful transcript is c… |
Example
Ask the agent in plain language, or call johnvc--YoutubeTranscripts with arguments such as:
{
"youtube_url": "https://www.youtube.com/watch?v=1UL8iZJD38U",
"languages": [
"en"
],
"output_formats": []
}
Output
JSON rows with these fields:
| Field | Type | Description |
|---|---|---|
url |
string | The original YouTube video URL that was processed. |
video_id |
string | The YouTube video ID (11 characters). |
language |
string | The transcript language name (e.g., 'English', 'English (auto-generated)'). |
language_code |
string | The ISO language code of the transcript (e.g., 'en', 'es', 'fr'). |
is_generated |
boolean | True if the transcript was auto-generated by YouTube, false if manually created. |
is_translatable |
boolean | Whether the transcript can be translated to other languages. May be null if not available. |
translation_languages |
array | List of language codes this transcript can be translated to. May be null if not available. |
total_seconds |
number | Total duration of the video in seconds, derived from the last transcript snippet. |
timestamped |
array | Array of transcript snippets with timestamps. Each snippet has text, start (seconds), and duration (seconds). |
non_timestamped |
string | Full transcript text without timestamps: all snippets concatenated. Use this for LLM input or readable export. |
timestamp |
string | ISO-format timestamp of when the video was processed. |
success |
boolean | True if the video was successfully transcribed, false if processing failed. |
cached |
boolean | True if this transcript came from the cache instead of a fresh fetch from YouTube. See fetched_at for its age; cached copies are at most 90 days old. Also present on the error row of a video with no usable captions: tru… |
fetched_at |
string | ISO-format UTC timestamp of when the transcript was fetched from YouTube. Equal to the run time for fresh rows; earlier for cached rows. On the error row of a video with no usable captions, it is when YouTube was last c… |
cache_age_days |
integer | Whole days between fetched_at and this run. Present only on cached rows. |
error |
boolean | True if an error occurred during processing; absent or false on success. |
error_message |
string | Human-readable description of the error, if any. |
error_type |
string | Type/class name of the error (e.g., 'ValueError', 'ValidationError', 'RequestBlocked'). A video with no usable captions reports 'NoTranscriptFound' (captions exist, but not in the requested languages or type; available_… |
is_permanent_error |
boolean | Present on the error row of a video whose transcript could not be fetched. True when asking again for the same video and languages cannot help: no usable captions, captions disabled, or the video is unavailable or age-r… |
is_blocking_error |
boolean | Present on the error row of a video whose transcript could not be fetched. True when YouTube refused the request itself, which is temporary: the same video may succeed in a later run. Always false when is_permanent_erro… |
retry_attempts |
integer | Present on the error row of a video that ended in a permanent or a blocking error: how many fetch attempts were made before the row was written. 0 when a 'no captions' answer came from the cache. |
translated_to |
string | ISO 639-1 language code the transcript was translated into. Only present when the translate_to input was set and translation succeeded. |
srt |
string | Transcript formatted as a SubRip (.srt) file. Present when 'srt' is included in output_formats. |
vtt |
string | Transcript formatted as a WebVTT (.vtt) file. Present when 'vtt' is included in output_formats. |
text |
string | Transcript as plain text, snippets joined with newlines (TextFormatter output). Present when 'text' is included in output_formats. |
list_only |
boolean | True if this dataset item was produced by a list_only=true run. The available_transcripts field will be populated and no transcript was actually fetched. |
available_transcripts |
array | Only present in list_only mode. Each entry describes one transcript variant YouTube exposes for this video: language, language_code, is_generated, is_translatable, translation_languages. One exception: the error row of… |
source_type |
string | Human-readable mirror of is_generated. "Manual" for human-created captions, "Auto-generated" for YouTube ASR (and for any translated transcript, since YouTube serves translations through the auto-gen pipeline). |
snippet_count |
integer | Number of timestamped snippets in the transcript (equal to len(timestamped)). |
duration_human |
string | Total duration formatted as a short human-readable string, e.g. "1h 19m 42s", "2m 7s", or "18s". See total_seconds for the raw float value. |
channel_name |
string | Name of the YouTube channel that uploaded the video. Populated when include_metadata is true. |
channel_id |
string | Stable YouTube channel ID (e.g., UC4QobU6STFB0P71PMvOGN5A). |
channel_url |
string | Direct URL to the YouTube channel. |
view_count |
integer | Total view count at the moment metadata was extracted. |
like_count |
integer | Total like count at the moment metadata was extracted. YouTube hides this for some videos; field is omitted when unavailable. |
video_duration_seconds |
integer | Video duration as reported by YouTube's player. Distinct from total_seconds (which is derived from the last transcript snippet); the two can differ slightly. |
upload_date |
string | Date the video was uploaded, formatted as ISO YYYY-MM-DD. |
thumbnail_url |
string | URL to YouTube's default thumbnail for the video. |
tags |
array | Tags the uploader assigned to the video. |
categories |
array | YouTube category labels for the video (e.g. "Film & Animation"). |
availability |
string | YouTube availability status, typically one of "public", "unlisted", "private", "premium_only", "subscriber_only", "needs_auth", "public_listed_unlisted". |
was_live |
boolean | True if the video was originally a livestream (now a replay). |
metadata_error |
string | If include_metadata was requested but metadata extraction failed, this field carries the error reason. The transcript fields are still populated. |
metadata_source |
string | Where the metadata fields came from: innertube (free, from the same YouTube response that carries the captions) or yt-dlp (include_extended_metadata, adds like_count, upload_date, categories, availability). Absent when… |
translation_error |
string | Set when translate_to was requested but the translation could not be delivered (the language is not offered, or YouTube throttled the translated caption download). The item then carries the original-language transcript… |
result_type |
string | Set to "channel_video" on rows produced by channel listing (the 'channel' input). Absent on transcript rows. |
source_channel |
string | The channel reference (name, @handle, ID, or URL) from the input that produced this row. Present on channel_video rows and channel listing errors. |
position |
integer | 1-based position of the video within its channel listing, newest first. Present on channel_video rows. |
Data source and terms
YouTube's public captions and subtitles, collected by scraping. The server returns the data as published at the time of the call. You are responsible for using it in line with the source's terms and the laws that apply to you. AlphaOSINT is not affiliated with the source.
Links
- Actor on Apify: https://apify.com/johnvc/YoutubeTranscripts
- Docs page: https://alphaosint.com/mcp/youtube-transcripts/
- All AlphaOSINT MCP servers: index
youtube-transcripts MCP questions
How do I add youtube-transcripts MCP to Claude Code?
Run claude mcp add --transport http youtube-transcripts 'https://mcp.apify.com/?tools=johnvc/YoutubeTranscripts' in your terminal. Add --scope user to make it available in every project.
How do I add youtube-transcripts MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
What tools does youtube-transcripts MCP provide?
It exposes 5 tools, including get-actor-run, get-dataset-items, get-key-value-store-record, abort-actor-run, johnvc--YoutubeTranscripts.
How much does it cost to process a video?
Each successfully transcribed video is charged at $0.00275, which includes the timestamped and plain-text transcript plus basic video metadata.
Can I get extended metadata like like counts and upload dates?
Yes, by setting include_extended_metadata to true, you can retrieve additional fields such as like count, upload date, categories, and availability for an extra charge.
How do I authenticate with this server?
You can use Apify OAuth which opens on first use, or provide an API token via the Authorization header.
Can I translate the transcripts?
Yes, you can use the translate_to parameter with an ISO 639-1 code to translate the picked transcript into a different language.