append_generated_media_chunk
ChatGPTAppend the next base64 chunk to a chunked upload started with begin_generated_media_upload. Send chunks in order with seq starting at 0, at most 8000 base64 characters each; if a call returns a 409 conflict, resume from the expectedSeq in the error details. Intended for files generated in this conversation (charts, cards, edited images); user-attached files and files on the user's device are handled by attach_media (the in-chat upload widget, full quality). Chunks are limited to 8000 base64 characters because larger payloads are silently truncated by the chat channel.
attach_media
ChatGPTUse this tool when the user wants to post a photo, video, or PDF document from their device or attached in the chat; it preserves full quality, whereas reading and base64-encoding attached files inline does not. On widget-capable clients (claude.ai, ChatGPT) it opens an upload widget IN the conversation (drag-and-drop or file picker, multiple files at once for carousels); on clients without widget rendering it still works — the result's structuredContent carries a programmatic upload API (ticket, presignUrl, confirmUrl) you can drive with plain HTTP. Use this whenever the user wants to post media that is not already at a public URL: a photo attached in chat, a video on their computer or phone, a PDF for a LinkedIn document post. The file uploads directly to the user's Schedulala media library at full quality (images to 20MB, videos to 500MB, PDFs to 100MB); when the upload finishes, the media's url and type are ready to pass into create_post's mediaItems (a PDF becomes type 'pdf' — LinkedIn documents only). For an Instagram Reel, also set platformSettings.instagram.postType to 'reel'.
begin_generated_media_upload
ChatGPTBegin a chunked upload for an image YOU generated in this conversation whose base64 is too large to send inline (a chart, quote card, or edited image). Returns an uploadId; then stream the base64 in small chunks via append_generated_media_chunk and finish with finish_generated_media_upload. Images only, 3MB decoded max. Intended for files generated in this conversation (charts, cards, edited images); user-attached files and files on the user's device are handled by attach_media (the in-chat upload widget, full quality). Chunks are limited to 8000 base64 characters because larger payloads are silently truncated by the chat channel.
bulk_create_posts
ChatGPTCreate multiple posts in one request (max 25). Each entry has the same shape as create_post (content, platforms, mediaItems, scheduledFor, publishNow, status, brandId, externalId, platformSettings, firstComment/firstComments - see create_post for first-comment platforms and limits). WARNING: this CONSUMES POST QUOTA per successful post (not per request). Validation/account-resolution failures are reported per-index and do NOT consume quota. TikTok posts require platformSettings.tiktok.privacy (e.g. 'SELF_ONLY' or 'PUBLIC_TO_EVERYONE') AND platformSettings.tiktok.postMode: 'direct' (publish straight to the profile) or 'inbox' (send to the user's TikTok app inbox to finish manually). There is no postMode default; entries without it fail validation. For media from the user's device or chat attachments, call attach_media first (where available) to get urls. Returns per-index results plus a summary of succeeded/failed.
cancel_post
ChatGPTCancel or delete a post by id. Draft posts are permanently deleted; scheduled posts are cancelled so they will not publish. Posts that are already processing or posted cannot be cancelled.
connect_account
ChatGPTGet the link to connect a new social media account (instagram, facebook, threads, linkedin, tiktok, twitter, youtube, pinterest, bluesky, telegram, mastodon, google-business) to Schedulala. google-business uses OAuth via the dashboard/connect URL like the other OAuth platforms (Google consent screen, then pick business locations). The link opens the Schedulala dashboard where the user completes the platform's own sign-in in their browser; credentials and OAuth approval NEVER happen in this chat. Use when the user wants to add or reconnect a platform. After they finish, call list_accounts to confirm the new account is connected.
create_post
ChatGPTCreate and schedule a social media post across one or more platforms via Schedulala. Supported platforms: twitter, instagram, linkedin, facebook, youtube, bluesky, tiktok, telegram, threads, pinterest, mastodon, google-business. Scheduling options (pick exactly one): scheduledFor (ISO 8601, 5 min to 90 days out), publishNow: true (publish immediately), queue: true (auto-scheduling queue), or status: "draft" (save without scheduling). Character limits: twitter=280, bluesky=300, threads=500, mastodon=500, google-business=1500, instagram=2200, tiktok=2200, linkedin=3000, telegram=4096, facebook=63206. (YouTube uses a 100-char title + 5000-char description; Pinterest uses a 100-char title + 500-char description — supply these via platformSettings.) google-business supports postType STANDARD/EVENT/OFFER with CTA buttons via platformSettings; exactly one image, no video. The CTA is two FLAT string fields, not a nested object: platformSettings["google-business"] = { "callToAction": "LEARN_MORE", "actionUrl": "https://..." }. Valid callToAction values: BOOK, ORDER, SHOP, LEARN_MORE, SIGN_UP, GET_OFFER, CALL. actionUrl is required for every type except CALL. Media: provide mediaItems with public URLs. Multiple items = carousel (instagram, facebook, threads, tiktok, pinterest; mastodon up to 4 images). Instagram requires at least one media item; YouTube requires a video. If the user's photos/videos are on their device or attached in the chat, attach_media (where available) opens an in-chat upload widget that uploads them at full quality and reports the urls to use here; reading and base64-encoding attached files inline loses quality. If the user supplies a source media URL to attach (or the link may expire), call upload_media first and use the url it returns here. Per-platform content: pass platforms as objects with a content field to override text per platform, or use platformSettings. First Comment (auto-plug): pass firstComment (plain string) or firstComments ([{target, text}]) to auto-post a comment under the post ~15 seconds after it publishes - the standard "link in the first comment" workflow. Supported on 8 platforms: linkedin, facebook, instagram, youtube, twitter, threads, bluesky, mastodon. NOT supported on tiktok, pinterest, telegram, google-business (no comment API): selecting ONLY unsupported platforms with a first comment is a validation error; mixed selections publish the comment where supported and warn about the rest. Comments are text and links only, no media. Comment limits: instagram=1000 UTF-8 bytes, linkedin=1500, twitter=280, bluesky=300, threads=500, youtube=10000 characters; facebook and mastodon have no fixed limit. target 0 = the main post; targets above 0 address thread entries and only work with create_thread (twitter/threads/bluesky). A failed comment never fails the post - check get_post's platforms[].firstComments for per-account delivery status (posted with the comment url, failed, or skipped). TikTok: platformSettings.tiktok.privacy (e.g. 'PUBLIC_TO_EVERYONE', 'SELF_ONLY') AND platformSettings.tiktok.postMode are both required. postMode is 'direct' (publish straight to the profile) or 'inbox' (send to the user's TikTok app inbox to finish manually) - there is no default, so the post is rejected until the user has chosen one.
create_thread
ChatGPTCreate a thread — a chain of sequential posts — on twitter, threads, bluesky, or mastodon. Provide at least 2 ordered entries in 'posts' (max 25). Each entry has its own content, an optional single media item (mediaItems, max 1 per entry), and optional per-platform text via platformContent. Scheduling: scheduledFor (ISO 8601) or publishNow: true; omit both to schedule for now. WARNING: this CONSUMES POST QUOTA — one post per thread entry (a 5-entry thread uses 5 posts of quota). Thread platforms only: twitter, threads, bluesky, mastodon. Other platforms are rejected. Use platformSettings for thread-wide options (e.g. mastodon visibility). config (postDelay, continueOnError, maxRetries) is accepted for compatibility, but the per-platform chain delay is fixed and postDelay is not honored. First Comment (auto-plug): pass firstComment (plain string, comments the root post) or firstComments ([{target, text}]) to auto-post comments after the thread publishes. target 0 = entry 1 (the root); target i = entry i+1, so max target = number of entries - 1. Thread-targeted comments (target > 0) deliver on twitter/threads/bluesky only; on mastodon only the target-0 comment applies. Comments are text and links only. Limits: twitter=280, bluesky=300, threads=500 characters; mastodon has no fixed limit. A failed comment never fails the thread - check get_post's platforms[].firstComments for delivery status. The response includes the thread id and a single postId (the whole chain publishes from ONE post document) — use get_post on that postId for live status.
finish_generated_media_upload
ChatGPTFinish a chunked upload: the server assembles the chunks, validates the image, and hosts it on the Schedulala CDN, returning the url + media id for create_post. Call once every chunk has been sent via append_generated_media_chunk. Intended for files generated in this conversation (charts, cards, edited images); user-attached files and files on the user's device are handled by attach_media (the in-chat upload widget, full quality). Chunks are limited to 8000 base64 characters because larger payloads are silently truncated by the chat channel.
get_best_posting_times
ChatGPTGet optimal posting times derived from historical engagement: top hours and top days of week, plus full by-hour and by-day breakdowns. Supported on these platforms only: instagram, twitter, bluesky, threads, facebook, linkedin, pinterest, tiktok, youtube, mastodon (not telegram or google-business). Pass accountId to scope to one account, or platform to scope to a platform, or neither to merge across all analytics-enabled accounts. Returns an empty result when there isn't enough posting history yet.
get_follower_growth
ChatGPTGet current follower count, 7-day and 30-day growth, and a daily follower-count history. Supported on these platforms only: instagram, twitter, bluesky, threads, facebook, linkedin, pinterest, tiktok, youtube, mastodon (not telegram or google-business). Note: tiktok reports no follower counts (platform limitation). Pass accountId (from list_accounts) for one account, or OMIT it to get per-account blocks for ALL analytics-enabled accounts plus summed totals in one call (totals exclude YouTube per its API policy). Optional window in days (1-365, default 30). Returns nulls/empty history when no snapshots exist yet.
get_post
ChatGPTRetrieve a single post by its id, including per-platform status, post URLs, and any errors.
get_post_analytics
ChatGPTGet per-post analytics (likes, comments, replies, shares, saves, views, reach, impressions, quotes, reposts, bookmarks, and engagement rate), with cursor pagination. Analytics is supported on these platforms only: instagram, twitter, bluesky, threads, facebook, linkedin, pinterest, tiktok, youtube, mastodon (telegram and google-business have no post-level analytics). Pass accountId (from list_accounts) for one account, or OMIT it to merge the top posts across ALL analytics-enabled accounts in one call (merged results don't paginate — pass accountId to page one account). Unsupported fields come back null per platform. Returns a clean empty result when no analytics exist yet.
get_usage
ChatGPTGet the current API plan, post quota usage (used/limit/remaining/percent), connected account count, billing period, and rate limit for this API key.
get_video_retention
ChatGPTGet per-video audience retention for a YouTube video on a connected channel (youtube only): what percentage of viewers are still watching at each point of the video, plus performance relative to similar YouTube videos. Fetched live from the YouTube Analytics API with a 7 day cache; requires the channel to be connected with the analytics scope. Data lags about 48 hours and needs enough watch activity to exist.
get_video_transcript
ChatGPTGet the caption tracks and parsed transcript of a YouTube video on a connected channel (youtube only). Useful for repurposing video content into posts on other platforms. videoId is the YouTube video id (from get_post_analytics platformPostId or the watch URL).
list_accounts
ChatGPTList the connected social media accounts available to this API key. Each account includes its accountId (use when targeting a specific account in create_post), platform, username/displayName, brandId/brandName (workspace — for dashboard-connected accounts this is the dashboard workspace the account is assigned to; group accounts by it, but do NOT pass a dashboard brandId to create_post), and connection health (isConnected false + needsReconnect true = the token is dead; tell the user to reconnect that account in the Schedulala dashboard instead of posting to it). There is no separate brands endpoint — read brand ids from here.
list_feed_keywords
ChatGPTList the saved feed keywords (social-listening watch list) for a platform. Shared with the Schedulala dashboard. Platforms: bluesky, threads, mastodon.
list_media
ChatGPTList the user's recently uploaded media (dashboard Media library + API uploads): url, type (image/video), file name, size and upload date, ready to pass into create_post's mediaItems. Use this whenever the user references media they've already uploaded — "my latest upload", "the video I uploaded", "that reel from earlier" — instead of asking them for a URL. This is THE path for posting videos from chat: big files can't travel through the conversation, so the user uploads once on the dashboard Media page and you reference it from here.
list_mentions
ChatGPTList posts/comments where your connected account is mentioned, tagged, or quoted. Supported platforms: instagram (tags + mentions), threads, bluesky (mentions + quotes), mastodon (mentions, which on Mastodon includes replies). (Not available on facebook, youtube, or linkedin.) For bluesky, each item's platformData.replyRefs lets you reply_to_comment directly.
list_posts
ChatGPTList posts created through the Schedulala API. Filter by status (draft, scheduled, posted, failed, cancelled), platform, or brandId, and paginate with page/limit. Date-range filtering uses 'after'/'before' (ISO 8601) on scheduled time.
lookup_profile
ChatGPTLook up any PUBLIC Bluesky profile by handle: bio, follower counts, and recent posts (with reply targets for reply_to_comment). bluesky only.
retry_post
ChatGPTRetry the failed platforms of a post. By default retries every platform currently in a 'failed' state; pass platforms to retry only a subset. Optionally pass delay (seconds) to reschedule the retry instead of running it immediately. Errors if the post has no failed platforms.
search_feeds
ChatGPTSearch PUBLIC posts by keyword (social listening). Platforms: bluesky (sort latest|top, cursor pagination), threads, and mastodon. Threads keyword search has a Meta-imposed budget shared with the Schedulala dashboard - the response reports usage; prefer bluesky or mastodon for exploratory searching, both of which are free public reads. Mastodon searches the PUBLIC HASHTAG TIMELINE for a single-word query (chronological, every instance) and falls back to full-text search for multi-word queries - full-text only covers posts whose authors opted into indexing, so a thin result set there is not evidence the topic is quiet; the response reports which path ran. Each result includes its reply target so you can respond via reply_to_comment (bluesky: platformData.replyRefs; threads and mastodon: the result id as postId).
search_instagram_music
ChatGPTFind a song for an Instagram Reel, on an Instagram account connected with the new Instagram login (list_accounts shows supports.music: true for those accounts; others get an error saying to reconnect from the dashboard). Without query it returns songs popular on Instagram right now; with query it searches by song or artist. Returns up to 25 songs, each with audioId, title, artist and duration, plus a next cursor for more. The chosen song goes on create_post as platformSettings.instagram.music: pass its audioId, plus title, artist, durationMs and artworkUrl, with platformSettings.instagram.postType 'reel' and a single video. previewUrl is a short-lived link for listening to a song before choosing; it is for listening only and is not stored on the post.
show_analytics
ChatGPTShow follower growth and top-post analytics for one connected account as an interactive card. Use when the user asks for analytics, stats, follower growth, or performance. Pass accountId for a specific account (from list_accounts); omit to use the first connected account.
show_calendar
ChatGPTShow the user's posts as an interactive calendar: upcoming scheduled posts (with a Cancel action) and recently posted ones (with a View link). Use when the user asks to see their schedule, calendar, upcoming posts, past posts, or post history.
show_comparison
ChatGPTShow a side-by-side comparison of 2-5 entities across metrics. Use to compare how one post did on different platforms, two accounts, or two time periods. YOU supply the gathered metrics per entity (from get_post_analytics / get_follower_growth). The widget aligns the metric rows and highlights the per-row winner.
show_connections
ChatGPTShow the user's connected social accounts and plan usage as an interactive panel. Use when the user asks to see their connected accounts, connections, or which platforms are linked.
show_feeds
ChatGPTShow posts matching the user's monitored keywords (social listening) as an interactive feed with Reply actions and keyword chips. Use when the user asks what people are saying about a topic, to monitor keywords/mentions, or to see their listening feed. Supported on Bluesky, Threads and Mastodon. Pass platform (required); optionally q (a search term) and accountId. Omit q to use the first monitored keyword.
show_inbox
ChatGPTShow recent comments (or comments on a specific post) for one connected account as an interactive inbox with Hide and Reply actions. Use when the user asks to see, review, moderate, or hide comments. Pass platform (required); optionally accountId and postId. Pass flaggedIds (comment ids you judged negative/spam) to highlight them and enable a one-click Hide-flagged button.
show_leaderboard
ChatGPTShow a ranked list of an account's posts by a metric as an interactive leaderboard. Use for 'best/top posts', 'most-liked', 'highest engagement', 'best videos last month', etc. Pass accountId (required) plus an optional sort (engagement|likes|impressions|date, default engagement), limit (default 10), and title.
show_post_preview
ChatGPTRender an interactive, faithful preview of a post draft across the selected platforms (per-platform mockups + live character counts; supports threads, image grids, carousels, PDF docs, 9:16 reels/shorts, and polls). READ-ONLY — it does NOT schedule or post. Use it after drafting a post, before the user confirms, so they can see exactly how it will look on each platform; then call create_post to actually schedule. Pass the same draft you'd give create_post.
update_feed_keywords
ChatGPTAdd or remove a saved feed keyword (social-listening watch list, max 20 per platform, shared with the Schedulala dashboard). Platforms: bluesky, threads, mastodon.
update_post
ChatGPTUpdate a draft or scheduled API post (posted/processing posts cannot be edited). Send only the fields you want to change: content, platforms, mediaItems (replaces all media), scheduledFor, timezone, status (draft/scheduled), platformSettings (per-platform content via platformSettings.<platform>.content), or firstComments (replaces the auto-posted first comments; an EMPTY array clears them). Content is re-validated against platform rules; an invalid update is rejected. Changing scheduledFor on a live scheduled post re-dispatches it.
upload_media
ChatGPTHost an image or video on Schedulala's CDN and get back a stable URL plus a media id for posting. Two modes (provide exactly one of url / data): 1. url — a publicly reachable http(s) source (a web image, an expiring link, a generated image URL). The server downloads and validates it. Limits: images up to 20 MB (jpeg, png, webp; gif up to 15 MB), videos up to 50 MB (mp4, mov, webm). For videos LARGER than 50 MB, skip this tool and pass the public URL directly in create_post's mediaItems; media is fetched at publish time. 2. data — base64-encoded file contents, LAST RESORT for tiny generated images only. When the user attached a photo or video in the conversation, or has a file on their device, attach_media is the right path: chat channels truncate large base64 payloads (a 6MB photo becomes a 6KB thumbnail), while attach_media opens an upload widget in the conversation and preserves full quality for images up to 20MB and videos up to 500MB. If the base64 exceeds ~8000 characters, do not send it inline - use begin_generated_media_upload instead. Then pass the returned url into create_post's mediaItems, e.g. mediaItems: [{ type: "image", url: "<returned url>" }]. The real file type is detected from the file contents. Re-hosted media expires after 30 days if unused.
validate_post
ChatGPTValidate post content against each platform's rules (character limits, media counts, required fields) WITHOUT creating or scheduling anything. Supports all 12 platforms including google-business (1500 chars; postType STANDARD/EVENT/OFFER with CTA buttons via platformSettings; exactly one image, no video). Returns per-platform valid/invalid with errors and warnings.