analyze_content_video
ChatGPTStart a saved Kabo analysis of a public YouTube, Instagram or TikTok video URL. Propose the analysis to the user and call only after they confirm; it needs Kabo Pro. Top-level fields, no parameters. It does not use the daily data allowance. Next: poll get_analysis_run with the returned id.
check_video_plan
ChatGPTCheck the storyboard written for a project and get its estimated cost. Free; nothing is made or saved. Top-level. ok false: fix what problem names and check again. Tell the user quote_usd and get their go-ahead before make_video.
claim_credit_task
ChatGPTClaim one one-time task only after the user asks. Tasks: bind_instagram, bind_tiktok, bind_youtube, follow_instagram, follow_facebook. Call get_credit_tasks first. not_ready = account not connected or official page not opened via its link. Not for daily tasks.
collect_account_history
ChatGPTAccount profiles + paged posts; reserves backfill. Top-level. Uses daily quota.
collect_instagram_trend_candidates
ChatGPTFind public Instagram Reels for a topic in a time window and save a candidate set. Top-level fields, no parameters. Not for a known account (use instagram_resolve_profile). Uses daily allowance. Never call a partial result complete. Next: pick candidates, then enrich_instagram_shortlist with candidate_set_id and selected.
collect_tiktok_video_candidates
ChatGPTTikTok search per intent and sort, plus optional daily board (empty: day before). Top-level. Uses daily quota.
collect_youtube_recent_videos
ChatGPTYouTube searches, deduped, then video details in 50-id chunks. Top-level. Uses daily quota.
collect_youtube_trend_candidates
ChatGPTFind public YouTube videos for a topic query plan in a publication window and save a candidate set. Top-level fields, no parameters. Not for a known channel. Uses daily allowance. 100 quota units per search, so keep the plan small. Next: pick candidates, then enrich_youtube_shortlist with candidate_set_id and selected.
connect_public_account
ChatGPTLink public profile; uses data allowance, no OAuth/analysis/sync. Replace only after user confirms; current id/bindingVersion: list_connected_accounts.
creator_trends_get_instagram_audio_board
ChatGPTRead Kabo's latest published Instagram trending-audio board (a snapshot, not live data). Inputs in parameters. No allowance cost.
creator_trends_get_tiktok_sound_board
ChatGPTRead Kabo's latest published TikTok trending-sound board (a snapshot, not live data). Inputs in parameters. No allowance cost.
enrich_instagram_shortlist
ChatGPTRank and enrich an Instagram shortlist you selected. Top-level fields, no parameters. Send platform, ranking_mode, window {since, until}, candidate_set_id from collect_instagram_trend_candidates (two lanes: candidate_sets [{lane, candidate_set_id}]) and selected [{content_key, semantic_scores {topic_relevance, locale_fit, creator_transferability?} 0-100, lane?, relevance?}]. Uses daily allowance.
enrich_youtube_shortlist
ChatGPTEnrich a YouTube shortlist you selected with exact metrics and same-channel baselines. Top-level fields, no parameters. Send platform, ranking_mode, window {since, until}, candidate_set_id from collect_youtube_trend_candidates and selected [{content_key, semantic_scores {topic_relevance, locale_fit, creator_transferability?} 0-100, relevance?}]. Uses daily allowance.
get_analysis_run
ChatGPTRead a saved video analysis: status, progress, transcript, result. Top-level fields, no parameters. If pending, call again later; failed and cancelled are final and come back as a normal result, not a tool error. No allowance cost.
get_creator_profile
ChatGPTStored niche/diagnosis/review.
get_credit_balances
ChatGPTRegular credits (today's allowance: used, remaining, reset time) and Video credits (separate balance; mode free = not charged, paid = blocked at 0). video_credits null = unreadable, not 0. Spends nothing.
get_credit_tasks
ChatGPTCredit task wall. Daily tasks (analyze a video, make a script, generate a video) pay out by themselves when done; never ask the user to claim them. One-time tasks wait for claim_credit_task. Show official_pages urls as named links; opening one shows a continue page.
get_for_you_videos
ChatGPTFor you: on-niche public videos by views (24h/7d/30d). Free. None today: starts one run (1/day), generating; retry after retry_after_seconds. not_eligible/failed say why. Unread 3 days pauses refresh.
get_kabo_info
ChatGPTKabo identity/scopes; no credits. name/image need profile; email needs email scope.
get_library_video
ChatGPTRead one saved video from the user's Kabo library by library_item_id: its details, saved analysis and linked scripts. Read-only. For the full analysis call get_analysis_run(analysisId); for a script call get_script.
get_remake_video
ChatGPTRead a video job. Top-level. project true: handles, revisions (each video.download is a final MP4) and the last step. producer true: the producer's state; at needs_reply relay message to the user and send their answer with reply_to_video. Links expire; call again to re-sign. cancel true stops the job's running work (spend so far stays charged). No quota.
get_research_capabilities
ChatGPTList Kabo's research data tools and whether each is ready for this account. Send {}. No allowance cost. ready is false while Kabo's own data source for that tool is unavailable.
get_research_data_download
ChatGPTGet a new signed link for a synchronous data result. Top-level fields, no parameters. Send its envelope's evidence_ref. No allowance cost. The link is valid for 10 minutes. It replays the recorded envelope, so sha256 and bytes can differ from an earlier link; media links (video_url, media_url) are not kept and come back null; YouTube comments get a newly issued next_cursor; lookup_instagram_post returns the route envelope. For jobs call get_research_data_job again.
get_research_data_job
ChatGPTRead a public_video_* data job. Top-level fields, no parameters. Send job_id and wait_s; if queued or running, call this again. failed and cancelled are terminal and come back as a normal result, not a tool error: check status and error_code. No allowance cost. Next: read listed artifacts with read_research_data_artifacts.
get_script
ChatGPTSaved script text/version.
get_social_connection
ChatGPTOfficial connection status/profile/scopes/sync time; read only.
instagram_baseline_search_music
ChatGPTSearch Instagram music and audio by keyword. Inputs in parameters. Uses daily allowance.
instagram_baseline_search_reels
ChatGPTSearch public Instagram Reels by keyword; not for account lookup. Inputs in parameters. Uses daily allowance.
instagram_get_transcript
ChatGPTRead the transcript of one public Instagram Reel by URL. Inputs in parameters. Uses daily allowance.
instagram_list_posts
ChatGPTRead a public Instagram account's feed posts by handle. Inputs in parameters. No allowance cost. Reels: instagram_list_reels. Send cursor back as max_id to continue.
instagram_list_reels
ChatGPTRead a public Instagram account's Reels by handle or user_id. Inputs in parameters. Uses daily allowance. Send cursor back as max_id to continue.
instagram_resolve_profile
ChatGPTRead a public Instagram profile (bio, counts, up to 6 recent media) by handle. Inputs in parameters. No allowance cost. For a known account start here, then instagram_list_reels or instagram_list_posts.
instagram_search_accounts
ChatGPTFind Instagram accounts by username fragment. Preferred over instagram_search_profiles. Inputs in parameters. No allowance cost.
instagram_search_profiles
ChatGPTFind Instagram accounts by free-text keyword. Inputs in parameters. No allowance cost. Use only when instagram_search_accounts finds nothing.
list_connected_accounts
ChatGPTLinked creator accounts. OAuth status: get_social_connection.
list_library_videos
ChatGPTLibrary summaries. Saved survives disconnect; platform: mine only. nextCursor pages.
list_scripts
ChatGPTScripts, including accountless; nextCursor pages. Text: get_script.
lookup_instagram_post
ChatGPTRead one public Instagram post or Reel as a fixed record: caption, owner, time, metrics. Top-level fields, no parameters. Send post as the user gave it, never a profile (use instagram_resolve_profile). Uses daily allowance. One lookup can use up to two allowance units.
lookup_tiktok_video
ChatGPTRead one public TikTok video as a fixed record: caption, author, time, metrics, sound. Top-level fields, no parameters. Send video as the user gave it, never a profile (use tiktok_list_account_posts). Uses daily allowance. One lookup can use up to two allowance units. Send the link rather than the bare id when you have it; some lookups only work from a link. Prefer this over tiktok_search_get_video.
official_page_render_page
ChatGPTOfficial platform help/policy page text by URL. parameters. No quota. Failure ≠ page silent.
prepare_creator_profile
ChatGPTFree review that builds a missing profile, inferring an unset niche. Then read get_creator_profile.
public_video_collect_public_metadata
ChatGPTRead a public video URL's metadata on any platform. Inputs in parameters. Deferred: returns a job_id. No allowance cost. Next: poll get_research_data_job until terminal; never re-call this to restart it.
public_video_fetch_transcript
ChatGPTFetch captions or a transcript for a public video URL. Inputs in parameters. Deferred: returns a job_id. Uses daily allowance. Up to 20 credits per call. Next: poll get_research_data_job until terminal; never re-call this to restart it.
public_video_measure_media_signals
ChatGPTMeasure shot cuts, brightness, silence, loudness, freeze frames and framing of a public video. Inputs in parameters. Deferred: returns a job_id. No allowance cost. Next: poll get_research_data_job until terminal; never re-call this to restart it.
read_research_data_artifacts
ChatGPTRead transcript or frame bytes listed in a data job's artifacts. Top-level fields, no parameters. Send job_id and object_refs from that job, never other paths or URLs. If any requested artifact's bytes are missing, the whole call fails: treat them as missing evidence. With delivery "download" each gets its own link and a missing one is marked unavailable instead. No allowance cost.
render_official_pages
ChatGPTRender 1-32 official help pages at once. Top-level. No quota.
reply_to_video
ChatGPTPass the user's answer, or a change they want to a delivered video, to the video producer of a job. May lead to paid steps the user asked for. Top-level. Then poll get_remake_video with project and producer true.
save_script
ChatGPTSave host text to Scripts. topic: 1-200 chars; script_text: 1-30000. No generation/credits.
start_skill_run
ChatGPTCall once as the first step of a bundled skill, with its name and the run directory's last path segment as run_id. Records the run once (same run_id is never recorded twice). credits_exhausted = stop and tell the user. Later tools in this run add no charge.
tiktok_discovery_get_industry_hashtags
ChatGPTRead TikTok Creative Center's trending-hashtag ranking, optionally by country, industry and period. Inputs in parameters. Uses daily allowance. Returns ranked hashtags with a short trend each, not videos. Rows carry industry_id and industry_name; to narrow to one industry, reuse an industry_id from an earlier result.
tiktok_discovery_get_search_insight_trend
ChatGPTRead the trend of one TikTok search insight by query_id from tiktok_discovery_list_search_insights. Inputs in parameters. Uses daily allowance.
tiktok_discovery_get_trending_hashtag_detail
ChatGPTRead the trend detail of one Creative Center hashtag by its numeric hashtag_id. Inputs in parameters. Uses daily allowance. Get hashtag_id from tiktok_discovery_get_industry_hashtags.
tiktok_discovery_list_search_insights
ChatGPTRead TikTok creator search insights: trending searches and content gaps. Inputs in parameters. Uses daily allowance.
tiktok_discovery_list_top_contents
ChatGPTRead TikTok Creative Center's daily top-content ranking. Inputs in parameters. Uses daily allowance.
tiktok_discovery_search_sounds
ChatGPTSearch TikTok sounds by keyword. Inputs in parameters. Uses daily allowance.
tiktok_discovery_search_videos
ChatGPTSearch public TikTok videos by keyword, with publish-time and sort filters. Inputs in parameters. Uses daily allowance.
tiktok_list_account_posts
ChatGPTRead a public TikTok account's recent videos (5–100) by handle. Default for account videos. Inputs in parameters. Uses daily allowance. Up to 10 provider requests per call.
tiktok_search_get_transcript
ChatGPTRead one TikTok video's transcript by URL (captions, else speech-to-text). Inputs in parameters. Uses daily allowance.
tiktok_search_get_video
ChatGPTRead one TikTok video by URL as a raw object, optionally with transcript. Inputs in parameters. Uses daily allowance. Prefer lookup_tiktok_video unless you need the transcript too.
tiktok_search_list_profile_videos
ChatGPTRead a TikTok account's videos by handle or user_id, sorted latest or popular. Inputs in parameters. Uses daily allowance.
update_content_preferences
ChatGPTSave the user's profile edits; no credits. First read get_creator_profile with effective:true: expected_version=preferences.version; basis=base.generatedAt (signature content). ids: sc/m_ (merge: 2-4 sc*; discard_unresolved: unresolvedEdits ref). text: excluded topic. null language/region follows the profile. Confirm guesses, remove, merge and reset with the user first. On conflict reread.
update_script
ChatGPTReplace text (1-30000 chars). expected_version: get_script; reread on conflict. No generation/credits.
upload_file
ChatGPTUser's local video (200 MB) or JPEG/PNG. upload goes to the plugin script, not chat; then file_id alone verifies. ref (kabo-file:id) fits any video-tool link field. No quota.
youtube_get_channel
ChatGPTRead a public YouTube channel's profile and stats by channel_id (UC…) or handle, not both. Inputs in parameters. Uses daily allowance.
youtube_get_video
ChatGPTRead public metadata and metrics for 1–50 YouTube videos by id. Inputs in parameters. Uses daily allowance.
youtube_list_channel_uploads
ChatGPTRead a YouTube channel's recent public uploads. Inputs in parameters. Uses daily allowance.
youtube_search_channels
ChatGPTFind public YouTube channels by keyword. Inputs in parameters. Uses daily allowance. Up to 100 quota units per call.
youtube_search_videos
ChatGPTSearch public YouTube videos by keyword and optional date window. Inputs in parameters. Uses daily allowance. Up to 100 quota units per call.