Skip to main content
Glama

LinkDigest — Xiaohongshu, Douyin, TikTok, YouTube to text

Read a social media post as text

digest_url
Read-only

Read what a social media link actually says. Works on Xiaohongshu, Douyin, Weibo, Hongguo 红果短剧 drama pages, Toutiao 今日头条 articles and videos, TikTok, YouTube, X, Instagram, Bilibili, podcast episodes (小宇宙, Apple Podcasts) and ordinary web pages. Returns the post's transcript with times, on-screen text, the text in every image, caption and counts, key points each backed by a quote checked against the post (key_point_support in json), and a receipt of what was read (coverage in json). Facebook and Spotify are not supported. Works without an API key for a first look (a summary, two a day per address); the full output needs a key — see check_balance. If the result names a job id, call again with job_id only. Example: url="https://v.douyin.com/abc/", ask="推荐了什么产品、多少钱?".

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "digest_url"
  }
}

Input Schema

NameRequiredDescriptionDefault
askNoA question to answer from the post, e.g. "what products and prices does it recommend?". Returns answer, found and the quotes it rests on, checked against the post; found is false when the post does not say — nothing is invented. +1 credit.
urlNoThe post link, or the share text that contains it. Required unless job_id is given.
depthNofull (default), or transcript for a long video: the whole transcript with a frame a minute, 1 credit per started 2 minutes and a longer length cap. script: the speech only (文案), 1 credit per started 2 minutes, no base. YouTube is always read in full.
formatNomarkdown (default) or json (structured: `coverage`, `key_point_support`, `answer`, and `links` — the links the post points to, with kind; None is opened).
job_idNoCollect a digest still running, instead of url (re-sending the url would start it again).
insightNoConsumer insight (消费者洞察): audience, claims, objections, purchase-intent cues, products/prices, questions — each a quote checked against the post and its top comments. +1 credit.
breakdownNoAlso take the post apart (爆款拆解): hook, timed beats, title formula, call to action, reusable template; quotes checked against the post. +1 credit.
partial_okNoRead the opening minutes the budget affords instead of refusing a long video; the digest then carries `partial`. Default false: a long video is refused for free.
translate_toNoAlso translate into this language (en, ja, zh-CN); the original text is kept. +1 credit.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed2 schema fields changed
    • changedInput schema / properties / depth / description
      Previous value: -"full (default), or transcript for a long video: the whole transcript with a frame a minute, 1 credit per started 2 minutes and a longer length cap. YouTube is always read in full."New value: +"full (default), or transcript for a long video: the whole transcript with a frame a minute, 1 credit per started 2 minutes and a longer length cap. script: the speech only (文案), 1 credit per started 2 minutes, no base. YouTube is always read in full."
    • changedInput schema / properties / depth / enum
      Previous value: -[
      -  "full",
      -  "transcript"
      -]New value: +[
      +  "full",
      +  "transcript",
      +  "script"
      +]
  2. Changed1 schema field changed
    • addedInput schema / properties / insight
      Added value: +{
      +  "description": "Consumer insight (消费者洞察): audience, claims, objections, purchase-intent cues, products/prices, questions — each a quote checked against the post and its top comments. +1 credit.",
      +  "type": "boolean"
      +}
  3. Changed8 schema fields changed
    • changedInput schema / properties / ask / description
      Previous value: -"A question to answer from the post, in any language: \"what products and prices does it recommend?\", \"does the speaker say X?\", \"what is step three?\". The answer comes with the quotes it rests on, each checked against the post by the server (`answer` in json: answer, found, support[]; an Answer section in markdown). `found` is false when the post does not address it — nothing is invented. Costs the digest's price plus 1 credit; the same question on the same url is served from cache for free."New value: +"A question to answer from the post, e.g. \"what products and prices does it recommend?\". Returns answer, found and the quotes it rests on, checked against the post; found is false when the post does not say — nothing is invented. +1 credit."
    • changedInput schema / properties / breakdown / description
      Previous value: -"Also take the post apart (爆款拆解): the hook in its first seconds, the structure as timed beats, the title formula, cover text, call to action, audience, and a reusable template to make your own version. Quotes are copied from the post and checked against it; engagement counts and hashtags come from the platform. Written in the post's language, or in translate_to if given. Shown as a section in markdown and as `breakdown` in json. Costs the digest's price plus 1 credit; a repeat is served from cache for free."New value: +"Also take the post apart (爆款拆解): hook, timed beats, title formula, call to action, reusable template; quotes checked against the post. +1 credit."
    • changedInput schema / properties / depth / description
      Previous value: -"How much of a video to read. full (default): the transcript plus frames sampled densely, 1 credit per started minute. transcript: transcript-first, for long videos — the whole transcript plus one frame a minute (or one per chapter), one credit per started 2 minutes, and a longer length cap (free 60 min, paid 120 min, against 10 and 30 for full). Coverage says the frames were sparse and `depth` in json says which read it was. When a full read would be partial, `partial.transcript_credits` is what this read of the whole video costs. A YouTube video read by Gemini watching it (YouTube blocks this server) has no transcript-only form: it is read in full and priced as a full read."New value: +"full (default), or transcript for a long video: the whole transcript with a frame a minute, 1 credit per started 2 minutes and a longer length cap. YouTube is always read in full."
    • changedInput schema / properties / format / description
      Previous value: -"markdown (default, best for reading) or json (structured). json adds `coverage` (seconds_transcribed, duration_seconds, largest_frame_gap_seconds, images_read, images_total, not_captured) and `key_point_support` ({ point_index, kind, t, image_index, quote, verified } per key point). Either format lists the links the post points to (a Links section; `links` in json: { url, found_in, label, kind }, from its caption, QR codes, WeChat's 阅读原文 and on-screen text). None is opened; kind marks payment codes and group invites."New value: +"markdown (default) or json (structured: `coverage`, `key_point_support`, `answer`, and `links` — the links the post points to, with kind; None is opened)."
    • changedInput schema / properties / job_id / description
      Previous value: -"Collect a digest already running. Pass the job id from a previous call instead of url. Use this rather than re-sending the url, which would start the work again."New value: +"Collect a digest still running, instead of url (re-sending the url would start it again)."
    • changedInput schema / properties / partial_ok / description
      Previous value: -"Read the opening minutes the budget affords instead of refusing a video over max_credits or the plan's length cap. Default false: a long video is refused and costs nothing (YouTube excepted: its length is known only after Gemini watches it, so it is read in full and billed at most max_credits). When true, the digest carries `partial` (read_seconds, duration_seconds, full_credits). An image post the budget does not cover whole reads its first images instead (6 a credit); its partial has 0 seconds, and image_count / images_read say how many."New value: +"Read the opening minutes the budget affords instead of refusing a long video; the digest then carries `partial`. Default false: a long video is refused for free."
    • changedInput schema / properties / translate_to / description
      Previous value: -"Also translate the digest into this language: an ISO 639-1 code, optionally with a region (en, ja, zh-CN). The original-language text is always returned; the translation is added alongside it under `translation` (title, caption, key points, transcript text, on-screen text, image descriptions), and as a Translation section at the end of the markdown. Machine translation. Costs the digest's price plus 1 credit; a repeat of the same url and language is served from cache for free."New value: +"Also translate into this language (en, ja, zh-CN); the original text is kept. +1 credit."
    • changedInput schema / properties / url / description
      Previous value: -"The post URL, including any share tokens. Required unless job_id is given."New value: +"The post link, or the share text that contains it. Required unless job_id is given."
  4. Changed1 schema field changed
    • addedInput schema / properties / ask
      Added value: +{
      +  "description": "A question to answer from the post, in any language: \"what products and prices does it recommend?\", \"does the speaker say X?\", \"what is step three?\". The answer comes with the quotes it rests on, each checked against the post by the server (`answer` in json: answer, found, support[]; an Answer section in markdown). `found` is false when the post does not address it — nothing is invented. Costs the digest's price plus 1 credit; the same question on the same url is served from cache for free.",
      +  "maxLength": 300,
      +  "type": "string"
      +}
  5. Changed3 schema fields changed
    • addedInput schema / properties / depth
      Added value: +{
      +  "description": "How much of a video to read. full (default): the transcript plus frames sampled densely, 1 credit per started minute. transcript: transcript-first, for long videos — the whole transcript plus one frame a minute (or one per chapter), one credit per started 2 minutes, and a longer length cap (free 60 min, paid 120 min, against 10 and 30 for full). Coverage says the frames were sparse and `depth` in json says which read it was. When a full read would be partial, `partial.transcript_credits` is what this read of the whole video costs. A YouTube video read by Gemini watching it (YouTube blocks this server) has no transcript-only form: it is read in full and priced as a full read.",
      +  "enum": [
      +    "full",
      +    "transcript"
      +  ],
      +  "type": "string"
      +}
    • changedInput schema / properties / format / description
      Previous value: -"markdown (default, best for reading) or json (structured)."New value: +"markdown (default, best for reading) or json (structured). json adds `coverage` (seconds_transcribed, duration_seconds, largest_frame_gap_seconds, images_read, images_total, not_captured) and `key_point_support` ({ point_index, kind, t, image_index, quote, verified } per key point). Either format lists the links the post points to (a Links section; `links` in json: { url, found_in, label, kind }, from its caption, QR codes, WeChat's 阅读原文 and on-screen text). None is opened; kind marks payment codes and group invites."
    • changedInput schema / properties / partial_ok / description
      Previous value: -"Read the opening minutes the budget affords instead of refusing a video over max_credits or the plan's length cap. Default false: a long video is refused and costs nothing (YouTube excepted: its length is known only after Gemini watches it, so it is read in full and billed at most max_credits). When true, the digest carries `partial` (read_seconds, duration_seconds, full_credits)."New value: +"Read the opening minutes the budget affords instead of refusing a video over max_credits or the plan's length cap. Default false: a long video is refused and costs nothing (YouTube excepted: its length is known only after Gemini watches it, so it is read in full and billed at most max_credits). When true, the digest carries `partial` (read_seconds, duration_seconds, full_credits). An image post the budget does not cover whole reads its first images instead (6 a credit); its partial has 0 seconds, and image_count / images_read say how many."
  6. Changed2 schema fields changed
    • changedInput schema / properties / url / description
      Previous value: -"The post URL, including any share tokens."New value: +"The post URL, including any share tokens. Required unless job_id is given."
    • changedInput schema / required
      Previous value: -[
      -  "url"
      -]New value: +[]
  7. Changed1 schema field changed
    • changedInput schema / properties / partial_ok / description
      Previous value: -"Read the opening minutes the budget affords instead of refusing a video over max_credits or the plan's length cap. Default false: a long video is refused and costs nothing. When true, the digest carries `partial` (read_seconds, duration_seconds, full_credits)."New value: +"Read the opening minutes the budget affords instead of refusing a video over max_credits or the plan's length cap. Default false: a long video is refused and costs nothing (YouTube excepted: its length is known only after Gemini watches it, so it is read in full and billed at most max_credits). When true, the digest carries `partial` (read_seconds, duration_seconds, full_credits)."
  8. Changed2 schema fields changed
    • addedInput schema / properties / breakdown
      Added value: +{
      +  "description": "Also take the post apart (爆款拆解): the hook in its first seconds, the structure as timed beats, the title formula, cover text, call to action, audience, and a reusable template to make your own version. Quotes are copied from the post and checked against it; engagement counts and hashtags come from the platform. Written in the post's language, or in translate_to if given. Shown as a section in markdown and as `breakdown` in json. Costs the digest's price plus 1 credit; a repeat is served from cache for free.",
      +  "type": "boolean"
      +}
    • changedInput schema / properties / translate_to / description
      Previous value: -"Also translate the digest into this language: an ISO 639-1 code, optionally with a region (en, ja, zh-CN). The original-language text is always returned; the translation is added alongside it under `translation` (title, caption, key points, transcript text, on-screen text, image descriptions) — visible in json format. Machine translation. Costs the digest's price plus 1 credit; a repeat of the same url and language is served from cache for free."New value: +"Also translate the digest into this language: an ISO 639-1 code, optionally with a region (en, ja, zh-CN). The original-language text is always returned; the translation is added alongside it under `translation` (title, caption, key points, transcript text, on-screen text, image descriptions), and as a Translation section at the end of the markdown. Machine translation. Costs the digest's price plus 1 credit; a repeat of the same url and language is served from cache for free."
  9. Changed1 schema field changed
    • addedInput schema / properties / partial_ok
      Added value: +{
      +  "description": "Read the opening minutes the budget affords instead of refusing a video over max_credits or the plan's length cap. Default false: a long video is refused and costs nothing. When true, the digest carries `partial` (read_seconds, duration_seconds, full_credits).",
      +  "type": "boolean"
      +}
  10. Changed1 schema field changed
    • addedInput schema / properties / translate_to
      Added value: +{
      +  "description": "Also translate the digest into this language: an ISO 639-1 code, optionally with a region (en, ja, zh-CN). The original-language text is always returned; the translation is added alongside it under `translation` (title, caption, key points, transcript text, on-screen text, image descriptions) — visible in json format. Machine translation. Costs the digest's price plus 1 credit; a repeat of the same url and language is served from cache for free.",
      +  "pattern": "^[a-z]{2}(-[A-Z]{2})?$",
      +  "type": "string"
      +}
  11. First observed

TDQS

A4.5/5.0
Behavior5/5

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

Annotations already declare readOnly/openWorld/not-idempotent, and the description adds substantial context beyond them: per-address free quota, credit costs for ask/insight/breakdown/translate_to/depth, the async job_id collection pattern with the warning that re-sending the url restarts the job, partial_ok behavior (long videos refused for free by default), and the presence of coverage and key_point_support receipts in the output. The re-send-restarts note is fully consistent with idempotentHint=false.

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

Conciseness4/5

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

Front-loaded with the core purpose, then platform list, then output shape, then cost/async mechanics, then an example — a sensible order. It is dense and the parenthetical asides (credit formulas, Chinese labels) make it feel cramped, but nearly every clause carries operational information rather than filler.

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

Completeness4/5

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

For a 9-parameter, zero-required, no-output-schema tool, the description covers output fields (coverage, key_point_support, transcript with times, on-screen and image text, counts), costs, failure modes (unsupported platforms, free-tier refusal), and the async continuation path. The main remaining gap is how long a job takes or how to poll, and it does not describe what the markdown default looks like versus json.

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

Parameters4/5

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

Schema coverage is 100%, so the schema alone already justifies a baseline of 3. The description goes further by explaining runtime semantics not in the schema: the url-or-job_id mutual exclusion and why re-sending url restarts work, and a concrete worked example pairing a Douyin URL with an ask question. Not every parameter is elaborated in prose, so it stops short of a 5.

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

Purpose5/5

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

Opens with a specific verb+resource ("Read what a social media link actually says") and then enumerates the exact source types it handles — Xiaohongshu, Douyin, Weibo, Hongguo, Toutiao, TikTok, YouTube, X, Instagram, Bilibili, podcasts, ordinary pages — plus explicit exclusions (Facebook, Spotify). This makes it distinguishable from compare_links/search_links/list_batches at a glance, since it is a single-link extraction rather than a batch or comparison tool.

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

Usage Guidelines4/5

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

Gives clear operating context: use the free no-key tier for a first look (two per day), use check_balance for a key to get full output, and follow up with job_id only when a result names a job id. Unsupported platforms are stated. It stops short of explicitly contrasting itself with compare_links or search_links, so alternatives guidance is implied rather than enumerated.

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

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.