SavedThat
Server Details
Search your saved Reels, TikToks and YouTube videos and creator libraries by what was said.
- Status
- Healthy
- Uptime
- 100.0% over the last 24 hours
- OAuth
- Works in Glama
- Last Tested
- Transport
- Streamable HTTP · MCP 2025-11-25
- URL
- Repository
- limerentt/savedthat-claude-plugin
- GitHub Stars
- 0
- ChatGPT
- –
- Claude
- –
TDQS
Scored across 11 tools
The set follows a clear two-scope pattern: user's own library (get_moment, get_my_transcript, list_my_saves, search_my_library) versus creator libraries (get_creator_moment, search_creator_videos, show_creators). The only friction is that the 'mine' side of the moment pair is named get_moment rather than get_my_moment, so an agent could momentarily misread the scope, though the descriptions resolve it.
Consistent snake_case verb_noun throughout (get_*, list_*, search_*, save_videos, mark_queue_done, show_*). The minor deviation is the asymmetric get_moment vs get_creator_moment and search_my_library vs search_creator_videos, where the 'my' marker appears only sometimes.
Eleven tools is well within the sweet spot, and each one maps to a distinct resource or action (save, list, search, fetch transcript, fetch moment, queue, creators, plans). Nothing feels redundant or padded.
Covers the core lifecycle: save videos, browse the library, search it, pull full transcripts or moment windows, manage the queue, plus discovery of creators and pricing. The main gap is no unsave/remove-from-library operation, which matters because saves count against a monthly quota and an agent cannot free space.
Publisher details
- Operator
- SavedThat · Publisher source
- Operator website
- https://savedthat.app · Publisher source
- Vendor relationship
- First-party · Publisher source
- Documentation
- https://savedthat.app/mcp · Publisher source
- Privacy policy
- Not provided
- Trust center
- Not available
- Restrictions
- Free plan has daily and monthly limits on searches of your own library; a paid plan (with a 7-day trial) removes them. Requires a SavedThat account. · Publisher source
Available Tools
11 toolsget_creator_momentGet a moment from a creator's videoARead-onlyIdempotentInspect
Returns the transcript around one moment (about a minute and a half) in a creator's video, given its url and timestamp, plus a link to the exact second. Returns an excerpt only, never a whole transcript. When found is false, reason says why: not_a_creator_video (the url is not in any creator library) or no_transcript_at_timestamp (the video is there, but that second has no transcript).
Inputs: url (required), at. Returns: at, url, found, range, +9 more
get_momentGet a focused transcript momentARead-onlyIdempotentInspect
Returns a focused window (about 2 minutes by default) of transcript around one moment in a video the user saved, enough context to act on an idea without a whole long video. Takes a video url, deep link or video id and a timestamp. Short videos come back in full. Source transcript marks spoken words, which can be quoted; visual means the video has no speech at all and the window holds the SavedThat indexer's scene and on-screen-text descriptions. When found is false, reason says why: not_processed_yet (still being transcribed), not_in_library (not saved by the user), transcription_failed (could not be transcribed), no_window_match (the timestamp is past the end of a video duration_seconds long).
Inputs: url (required), at, window_sec. Returns: at, url, full, found, +11 more
get_my_queueGet my queueARead-onlyIdempotentInspect
Lists the videos the user sent from the SavedThat app with its 'Send to' button, newest first: title, channel, url, the moment they picked (at_seconds), a deep link to that second and a queue_id for each entry.
Inputs: limit. Returns: items, total
get_my_transcriptGet the full transcript of a saved videoARead-onlyIdempotentInspect
Returns the entire transcript of one video the user saved. A long podcast can run to tens of thousands of words. When the video has no speech at all, source is visual and the text is the SavedThat indexer's scene and on-screen-text pass, not the creator's words. When found is false, reason says why: not_processed_yet (still being transcribed), not_in_library (not saved by the user), transcription_failed (could not be transcribed), no_transcript (indexed but holds no readable pass).
Inputs: url (required). Returns: full, found, reason, source, +1 more
list_my_savesList my recent savesARead-onlyIdempotentInspect
Lists the videos the user saved, newest first, with title, channel, platform, URL, saved date and whether the transcript is ready yet. Supports paging and a filter on title or channel, for browsing the library without a search query.
Inputs: limit, offset, platform, title_contains. Returns: saves, total, next_offset
mark_queue_doneMark a queued video doneAIdempotentInspect
Closes one entry in the user's queue, given its queue_id. The video stays in the user's library; only the queue entry is closed.
Inputs: queue_id (required). Returns: done
save_videosSave videos to my libraryAIdempotentInspect
Adds YouTube, Instagram or TikTok video links to the user's SavedThat library, where each one is transcribed and becomes searchable within a few minutes. Takes up to 25 links per call, counted against the same monthly save quota as saving in the app. Returns one result per link: saved, already_saved, invalid_url (not a supported video link), too_long (longer than the plan allows), limit_reached (the plan's save or library limit is used up; later links are not_attempted), or failed.
Inputs: urls (required). Returns: saved, results, not_saved, library_url, +2 more
search_creator_videosSearch a creator's videosARead-onlyIdempotentInspect
Searches everything one creator said across their transcribed videos, by meaning and in any language. Returns up to 5 distinct videos, each with its best moment, the exact quote, a timestamp and a link that opens the video at that second. The videos do not need to be in the user's own library. The creator is given by name, handle or slug. When candidates is empty, reason says why: creator_unknown (SavedThat has no library for that creator; available_creators lists the ones it has), creator_ambiguous (several creators matched, listed in did_you_mean), no_match (the creator's searched_videos transcribed videos do not cover it), search_unavailable or search_failed (a temporary SavedThat error).
Inputs: query (required), creator (required), limit. Returns: query, reason, creator, candidates, +3 more
search_my_librarySearch my saved libraryARead-onlyIdempotentInspect
Finds which of the user's own saved videos match a topic or a half-remembered detail. Returns up to 5 distinct videos rather than raw fragments, each with its single best moment, a timestamp, a match score and a source. Source transcript marks spoken words, which can be quoted; visual and metadata mark a scene or on-screen-text description written by the SavedThat indexer, not the creator's own words. When candidates is empty, reason says why: library_empty (nothing saved yet), no_match (the searched_videos saved videos hold nothing on this), search_unavailable or search_failed (a temporary SavedThat error).
Inputs: query (required), limit, channel, platform. Returns: query, reason, degraded, candidates, +1 more
show_creatorsShow creator librariesARead-onlyIdempotentInspect
Lists the creators whose videos SavedThat has already transcribed (for example Alex Hormozi, Andrew Huberman, Y Combinator), with niche, video count, hours and sample questions, rendered as an interactive picker. Optional filters by name or niche.
Inputs: niche, query. Returns: creators
show_plansShow SavedThat plansAInspect
Shows SavedThat's paid plans (Pro and Power, monthly or yearly) with current prices and a checkout or billing-portal link for each, rendered as an interactive card. Returns only links: nothing is charged by this tool.
Inputs: no named parameters. Returns: note, plans, channel, currency, +3 more
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Connectors
Transcribe, read and download public Reels, TikToks and YouTube videos, plus sourced creator data
Transcripts of TikTok, Instagram, YouTube, X, Facebook and RedNote videos, saved to your library.
Search, read, and organize your saves from X, Reddit, TikTok, Instagram, and more.
Get transcripts from YouTube, TikTok, X, Instagram and more - even when captions are off.
Related MCP Servers
- AlicenseNot gradedqualityBmaintenanceSearch within videos by dialogue, on-screen visuals, and recurring faces using natural language - connected with your favorite agent (like Claude, Codex, Hermes).60MIT
- AlicenseNot gradedqualityCmaintenanceEnables agents to research and analyze public short-form video content on Instagram Reels, TikTok, and YouTube Shorts for trend spotting, creator research, competitive briefs, and content ideation.1MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI agents to search a private, timestamped YouTube transcript library and retrieve cited passages with video links and exact timestamps. Supports hybrid vector and full-text search across entire channels so agents can answer from trusted sources rather than rewatching videos.1MIT
- AlicenseNot gradedqualityBmaintenanceExposes a locally built, searchable index of your saved Instagram reels — AI summaries, key points, transcripts, categories, and every extracted tool, place, product, and link — to any AI assistant over MCP. Enables chat-based search and retrieval across the whole collection without leaving the conversation.1Apache 2.0
Glama MCP Gateway
Add one secure layer between your agents and this server.