delete_automation
ChatGPTPermanently delete an Automation and its schedule. Call only when the creator explicitly asks to delete this specific Automation; to stop it temporarily, use save_automation with enabled false. A run already in progress still delivers its output, and content the Automation produced is kept. Returns the deleted ID and name.
delete_content
ChatGPTPermanently delete a Podcast, Episode, Video, Short, or Voiceover by ID. Deletion is irreversible: there is no undo or trash. A published Episode disappears from its Podcast feed, and deleting a Podcast deletes all of its Episodes. Call only on an explicit creator request that names the item; confirm with the creator first when the item is a Podcast or a published or scheduled Episode. Returns the ID, resolved kind, and, for an Episode, the publication state it had. An ID that no longer exists returns not found.
export_episode
ChatGPTProduce downloadable audio, video, or both from an existing Episode without regenerating its narration or publishing it. Reuses current files when possible; otherwise the new export replaces the current download for each requested format. Each file becomes available independently, and a failed replacement preserves the previous successful download. Poll get_content and read files for readiness. Reuse idempotency_key only when resending the same request after an uncertain result.
export_video
ChatGPTRender the playable stream, share video, and downloads of an existing Video, Short, or Voiceover from its saved content, for example after update_content changed captions or the watermark, or when get_content reports progress.render_required. Episodes use export_episode. Reuses the current render when one exists. It never regenerates narration and spends no Credits; when narration is out of date the error asks the creator to finish in Studio. Returns the ID, kind, outputs, and progress; poll get_content honoring progress.poll_after_seconds until outputs.video_stream is available or progress.status reads failed. Reuse idempotency_key only when resending the same request after an uncertain result.
generate_episode
ChatGPTGenerate editable Episode content from a prompt and optional public webpage, YouTube, or client-held plain-text Sources, or preserve a finished script. Use outputs to request audio, video, both, or [] for an editable draft. Omission preserves automatic export. Each requested file becomes available independently. Export saved drafts later with export_episode. Explicit creator requests need no extra confirmation. Ask before an agent-proposed generation because it consumes Credits. Episodes remain drafts unless auto_publish is true, which requires an explicit creator request and publishes only after generation and rendering succeed. Returns the ID, kind, draft publication state, progress, poll interval, applied Character settings, and the Studio URL of the draft workspace (get_content returns the stable /episodes address for the same Episode). Poll get_content with the returned ID.
generate_short
ChatGPTAlias of generate_video whose target_duration_seconds defaults to 30. Generate and render a Video from a prompt plus optional public webpage, YouTube, or client-held plain-text Sources, or voice a finished script without rewriting it. For target_duration_seconds, 30 or 60 seconds makes a Short; 180 or more makes a Video. background_music applies only to Shorts. style_id is required and must come from list_video_styles. Prompt generation uses character_ids for an ordered cast of up to four Characters, or character_id for one narrator. Script speakers resolve from [Character Name] labels or speaker_characters assignments. Explicit creator requests need no extra confirmation. Ask before an agent-proposed generation because it consumes Credits. Returns the ID, kind, Studio URL, progress, poll interval, and applied settings. Poll get_content with the returned ID. Reuse idempotency_key with identical input after an uncertain response. After a rejection before generation starts, follow recovery.idempotency_key; a failed generation that already started requires a new key.
generate_video
ChatGPTGenerate and render a Video from a prompt plus optional public webpage, YouTube, or client-held plain-text Sources, or voice a finished script without rewriting it. For target_duration_seconds, 30 or 60 seconds makes a Short; 180 or more makes a Video. background_music applies only to Shorts. style_id is required and must come from list_video_styles. Prompt generation uses character_ids for an ordered cast of up to four Characters, or character_id for one narrator. Script speakers resolve from [Character Name] labels or speaker_characters assignments. Explicit creator requests need no extra confirmation. Ask before an agent-proposed generation because it consumes Credits. Returns the ID, kind, Studio URL, progress, poll interval, and applied settings. Poll get_content with the returned ID. Reuse idempotency_key with identical input after an uncertain response. After a rejection before generation starts, follow recovery.idempotency_key; a failed generation that already started requires a new key.
get_automation
ChatGPTGet one Automation by ID with the same profile list_automations returns, for example to confirm a save_automation change. Deleted Automations are not found.
get_content
ChatGPTGet a Podcast, Episode, Video, Short, or Voiceover by ID. Returns progress, Studio destination, direct audio and video streams for playback, download links, output states, and, for an Episode with include_transcript: true, the speaker-attributed transcript text (capped, with a truncated flag). progress.status reports production progress for every kind: read it, not content.phase or content.generation. For an Episode, content.status is its publication state: draft, published, or scheduled. Videos, Shorts, and Voiceovers return no thumbnail image. An output is usable when its status is available. For Episodes, files.audio and files.video report independently requested file readiness, current-content flags, download URLs and failures, including files ready before the whole request completes. A completed Episode whose files read not_requested plays in Studio but has no file yet: call export_episode to produce one. When progress.render_required is true, status reads completed but no output exists and none will appear on its own (for example after a caption or watermark edit): call export_video for a Video, Short, or Voiceover, or export_episode for an Episode, instead of polling. progress.skipped_sources lists Sources generation left out. When content.visual_limit_seconds is set, a trial made visuals only that far. Narration runs longer, and every render ends there until the creator starts their plan and finishes the Video in Studio. If the client cannot show the player, present the available stream as a Listen or Watch link. This tool never starts generation or rendering.
get_insights
ChatGPTRead how the whole library performed over a period, with the same numbers as Studio Insights: total plays, watch time, items published, and Channels posted to, each against the previous period of equal length; plays and posts per source (podcast feeds, the Jellypod player, each Channel); the top 5 items by plays; and up to three suggested next steps with Studio links. Use period last-week for "how did last week go". complete false means a source has not finished its first capture, so numbers are a floor; as_of is when the numbers were captured. Per-podcast countries, apps, devices, charts, and exports stay in Studio at studio_url.
list_automations
ChatGPTList saved Automations, newest first, each with its profile: name, prompt preview, enabled state, pause reason, trigger type, schedule or inbound address, output type and settings, and the linked Podcast or narrator Character. Prompts longer than 300 characters are cut (prompt_truncated); call get_automation for the full prompt. Automations with identical full prompts share a prompt_fingerprint, so duplicates are visible. Follow nextCursor while hasMore. Default 20 rows, maximum 50.
list_characters
ChatGPTList saved Characters with their profile and Voice details. Backstory and personality longer than 300 characters are cut (profile_truncated); pass include_full_profile for the full text. Follow nextCursor while hasMore. Default 20 rows, maximum 50.
list_content
ChatGPTBrowse Podcasts, Episodes, Videos, Shorts, and existing Voiceovers together, newest first. Optionally filter by kind, title, generation state, Episode publication state, or parent Podcast. Returns 20 rows per page unless limit is set (up to 50). Follow nextCursor while hasMore. A completed generation_state does not prove outputs exist; get_content reports them and progress.render_required.
list_video_styles
ChatGPTList built-in Video style IDs, names, descriptions, and landscape and portrait sample images of each style. A sample shows the style, never a preview of any Content.
list_voices
ChatGPTList built-in Voices with stable numeric IDs and descriptive attributes. Use query for plain-text search across description, accent, and style tags, or accent for an exact accent. Use a returned ID when choosing a Voice for a Character, and describe Voices to the creator by name and attributes. Follow nextCursor while hasMore. Default 20 rows, maximum 50.
save_automation
ChatGPTCreate an Automation when id is omitted, which requires trigger_type, output_type, and podcast_id (Episode) or character_id (Video, Short), plus a prompt and schedule for scheduled runs. Update when id is supplied: pass only the fields to change, such as prompt, name, schedule, enabled, the linked Podcast or Character, or output settings. Trigger type and output type cannot change. Scheduled runs can publish Episodes when auto_publish is explicitly requested; inbound_email runs accept mail sent to the returned address. Each Automation run consumes Credits like a creator-started generation. Returns only the ID, kind, operation, and saved values.
save_character
ChatGPTCreate a Character when id is omitted, or update name, title, backstory, and personality when id is supplied. Voice changes and archiving require Studio. Returns only the affected ID, kind, operation, and saved values.
save_podcast
ChatGPTCreate a Podcast when id is omitted, which requires title and character_ids, or update descriptive Podcast metadata (including author and Apple Podcasts categories) when id is supplied. Visibility controls Podcast discovery and never publishes or unpublishes an Episode. Returns only the affected ID, kind, operation, and saved values.
set_episode_publication
ChatGPTPublish an Episode now, schedule or reschedule publication, or return it to draft. When the Episode has finished generating but its audio is missing or older than its latest edits, Jellypod renders it first and returns publication_state publishing or scheduling; poll get_content until content.status reads published or scheduled. Requires explicit creator authorization for the Episode and action. Publishing exposes it only when its Podcast is not private; Episode publication does not override Podcast visibility. Unpublishing cannot recall downloaded copies. Publish and schedule receipts carry the public Episode URL and the Podcast RSS feed URL, both null while the Podcast is private. An Episode still generating or rendering is refused with a poll instruction. An Episode whose generation failed, or whose narration is out of date or missing, is refused with error.studio_url, because fixing it needs Jellypod Studio.
submit_feedback
ChatGPTSend a support report to Jellypod only when the creator explicitly asks to submit feedback or report a connector problem to Jellypod. A complaint or tool error alone is not permission to send a report. Send only the creator-approved goal, relevant tool, and observed and expected behavior. No account identity, plan, or organization ID is attached. Tell the creator after submitting. Include only a concise description of the Jellypod connector workflow. Never include conversation transcripts, creator content, secrets, or unrelated host or model feedback.
update_content
ChatGPTUpdate at least one of title, description, explicit, captions_enabled, caption_style_id, or watermark on an Episode, Video, Short, or Voiceover by ID. Pass them as top-level arguments beside id, not wrapped in another object. Video, Short, and Voiceover accept title but not description or explicit. Caption and watermark fields override the default (the Podcast default for an Episode, the Organization default for a Video); null returns a field to that default, and watermark only moves an existing watermark. Jellypod resolves the kind. Scripts, narration, cast, Sources, visuals, timeline, and new watermark images remain Studio edits. Returns the ID, resolved kind, and saved values. A caption or watermark change clears the current render, so rerender in the result names the outputs that are now unavailable and the export tool (export_video, or export_episode for an Episode) that renders them again; rerender is absent when the settings already held these values.