← ChatTubeCONTENT HISTORY
WHAT CHANGED
Update to ChatTube
Snapshot Nov 18, 2026 · 06:17 UTC · version 1.0.0
Collection source: downloaded plugin package.
First saved snapshot
No earlier snapshot is available to establish a change.
Save ChatTube and check future changes in Following.
Up to 3 plugins free · no card required.
Compare saved observations
Download comparison JSONFull technical diff · 0 changed fields
Full snapshot data
{
"description": "Search YouTube, summarize videos, extract key points, and answer questions about video content using ChatTube. Handle ChatTube tool results, including exhausted AI credits and account-entitlements guidance.",
"included_files": [],
"name": "chattube",
"skill_md_contents": "---\nname: chattube\ndescription: Search YouTube, summarize videos, extract key points, and answer questions about video content using ChatTube. Handle ChatTube tool results, including exhausted AI credits and account-entitlements guidance.\n---\n\n# ChatTube workflow\n\nUse the connected ChatTube tools for the user's YouTube request. Ask for a missing\nvideo URL or other required input when it cannot be determined from the conversation.\nUse `youtube_video_transcript` for summaries and questions about video content;\nuse the relevant search, video, channel, playlist, or comment tools for other requests.\nCall `get_profile` only when the user asks about their ChatTube account or credits.\n\nInspect each tool result before calling another tool or composing the answer.\n\n## Before requesting a transcript\n\nCheck the user's intended output before calling `youtube_video_transcript`. If\nthey ask to download, export, transcribe, or reproduce a complete transcript or\nsubtitle file, including SRT, VTT, JSON, or plain text, do not call the tool.\nExplain that complete transcripts and subtitle exports are unavailable due to\ncopyright and content licensing restrictions. Offer a summary, key points,\nchapters, or answers about the video's content instead. Use the transcript tool\nonly for these transformative tasks.\n\n## AI credit usage\n\nEach successful ChatTube tool call, except `get_profile`, costs 1 AI credit,\nincluding calls for subsequent pages and successful calls with no results.\nInternal API requests and retries within a tool call do not add charges. Failed\ncalls, `credits_exhausted` results, and `get_profile` are free.\n\n## When AI credits are exhausted\n\nA ChatTube tool returns `structuredContent.status: \"credits_exhausted\"`\nand `structuredContent.entitlementsUrl`. This is the end of the current ChatTube\nworkflow, not a temporary API failure or an empty search result.\n\n1. End the current ChatTube workflow. Do not retry the credit-consuming ChatTube\n call or make further credit-consuming ChatTube calls for this workflow.\n2. Briefly explain in the user's language that they have used up their ChatTube AI\n credits. Use a friendly, conversational tone and include a clickable\n account-entitlements link using `entitlementsUrl`.\n If the URL is missing, use `https://chattube.io/entitlements`.\n3. Keep the ChatTube credit notice brief. Do not claim that unavailable ChatTube\n results were retrieved, or promote an upgrade, subscription, purchase, pricing,\n or checkout flow.\n\nExample response in English:\n\n> You've used up your ChatTube AI credits. You can take a look at your [account entitlements](https://chattube.io/entitlements) for details about credits and plan benefits.\n\nAlso apply this branch to a legacy ChatTube tool result that explicitly says all\nChatTube AI credits have been used. Do not infer exhausted credits from unrelated\ntechnical errors, a missing login, or an empty result.\n\nThis status applies only to ChatTube's AI credits, not to the conversation as a\nwhole. It does not restrict other plugins, tools, information sources, or later\nuser requests. Handle subsequent non-ChatTube requests normally without repeating\nthe credit notice. A new ChatTube call that reports exhausted credits should\nreceive the same notice and entitlements link.\n\n## Pagination\n\nFor tools that support pagination, omit `pageToken` on the first call. Fetch\nanother page only when needed to meet the user's requested number of results or\nscope. Pass the returned `nextPageToken` unchanged as `pageToken` to the same tool,\nkeeping the same resource, query, filters, sort order, and locale settings.\nNever invent a token or reuse one for a different tool or query.\n\nStop when the requested coverage is satisfied, `hasMore` is false, or\n`nextPageToken` is null. Do not automatically retrieve every available page;\neach successful page costs another AI credit.\n\n## Transcript language\n\nPass `language` to `youtube_video_transcript` only when the user explicitly asks\nfor a particular subtitle language. Otherwise omit it and let the server choose\nthe default track, preferring English when available. The user's requested reply\nlanguage does not by itself select a subtitle language. `lang` only localizes API\nmetadata; it does not select or translate the transcript.\n\nIf `resolvedTranscript.matchedRequestedLanguage` is false, briefly explain the\nfallback using `resolvedTranscript.selectedLanguage` and `resolvedTranscript.message`.\nUse the returned transcript for the analysis rather than repeating the same\nrequest to obtain the unavailable language. Reply in the user's requested language\nregardless of the subtitle language.\n\n## Timestamp links\n\nWhen summarizing a video, listing key points, generating chapters, or answering\nquestions about its content, add clickable timestamp links next to the relevant\nsentences or bullet points whenever the returned data provides reliable timing.\nFor answers covering multiple videos, link each point to the video it comes from.\n\nUse the start time of the transcript segment or chapter that supports the point.\nBuild links as `https://www.youtube.com/watch?v=VIDEO_ID&t=SECONDSs`, replacing\n`VIDEO_ID` with the actual video ID and `SECONDS` with the start time in whole\nseconds, rounded down. If a JSON3 segment provides `tStartMs`, divide it by 1,000\nbefore rounding. Use `MM:SS` as the link label, or `H:MM:SS` for times of an hour\nor more.\n\nExample for a supporting segment starting at 135 seconds:\n\n> The speaker introduces the main idea [02:15](https://www.youtube.com/watch?v=VIDEO_ID&t=135s).\n\nKeep links close to the content they support instead of collecting them in a\nseparate list. Do not invent or estimate timestamps from the topic, text length,\nor video duration. If reliable timing is unavailable, omit the timestamp and use\na regular video link when available.\n\n## Other results\n\nFor successful results, answer from the returned data and state any missing\ninformation. Do not reproduce complete transcripts, expose raw subtitle content\nor subtitle download URLs, or reconstruct a transcript through repeated excerpts.\nIf login is required, let the user complete ChatTube sign-in. For other tool errors,\nexplain the failure without claiming the user's credits are exhausted.\n"
}SHA-256 of public snapshot: a0d66bda074155d8621d7c0b56987bba079bcd25e1cca894eb85e6d564c87465