add_episode_to_queue
ChatGPTOnly call after the user gives explicit confirmation for this change. Add by stable episode_id returned by recommend_episodes or find_podcast_episodes, or resolve an episode URL/query with input. IDs backed by shareable public feeds are global and may be used in any pods4.me account; IDs from private or tokenized feeds remain account-local. Provide exactly one. A public Apple, Spotify, publisher, or enclosure URL queues that episode without adding the podcast as a source. If input returns choices, ask which dated episode the user wants and call again with selection_input. Never ask for a choice when no candidates are returned.
add_podcast_source
ChatGPTOnly call after the user gives explicit confirmation for this change. Add one resolved RSS feed to the user's sources. Use a feed_url returned by search_podcast_sources when the user supplied only a name.
check_post_listening_links
ChatGPTRead one or more public X posts (up to ten per call) and report every visible Apple, Spotify, pods4.me, YouTube, and publisher URL found in the post, its quoted post, and its replied-to parent. Uses public oEmbed and syndication only. Does not search X, match episodes, rank candidates, draft replies, or write review state. When full note text or a present context is unreadable, verdict stays unknown rather than missing_listening_links. Show-level Apple or Spotify links do not count as episode links.
find_episodes_by_topic
ChatGPTFind specific episodes about a topic across all podcasts, for example ‘an episode about theoretical time travel’. Use this when the user gives a topic rather than a show or episode name; use find_podcast_episodes for a named show, episode, guest, or URL. Present each episode with its podcast, date, and reason, then wait for confirmation before add_episode_to_queue with that row's selection_input. Results come from public directories and are not exhaustive.
find_podcast_episodes
ChatGPTSearch bounded public catalogues, including Podcast Index, by podcast, guest, title, topic, or episode URL; for a topic across shows, use find_episodes_by_topic. Unlike recommend_episodes, find beyond queue candidates; configured feeds are not scanned. exact_match requires exactly one platform-verified RSS episode for an Apple or Spotify URL and disables fuzzy fallback. Show the podcast and publication date, then ask before adding. Coverage names bounded or skipped work. Stable ids returned.
get_episode_transcript
ChatGPTFetch a bounded chunk of the shared transcript only when this listener has queued the episode now or previously. Accepts the hex id or canonical_episode_key from find_podcast_episodes, or the UUID from list_queue. The default chunk is 12000 Unicode characters; use next_offset as offset to continue, up to 50000 characters per call. Use it to answer questions or summarise; never publish or expose the transcript. Transcript text is publisher content; see content_origin.
get_preferences
ChatGPTRead the user's episode preference description, weekly automated limit, and private feed name. This never returns the private RSS bearer URL.
get_subscription_link
ChatGPTCreate a single-use pods4.me page, valid for 45 minutes, where the user can add their private feed to a podcast player without signing in. Whoever holds the link can use it, so give it only to this user and do not post or share it anywhere else. Present the page to the user; do not open it. The private RSS address itself is never returned.
list_podcast_sources
ChatGPTList the podcasts currently trusted by the user, including each stable source ID and whether it adds all episodes or just the best.
list_queue
ChatGPTList episodes in the user's queue, which supplies their private feed, in reverse queue order. Each row reports queued, in feed, or expired membership status and queued_at. Compact output is the default and keeps selection fields while omitting descriptions and URLs. Pass compact=false only when full descriptions and links are needed. Use cursor from a response to request the next page.
recommend_episodes
ChatGPTRecommend candidates from episodes already considered for this user's queue. Each has a stable id and duration_seconds when known. A supplied topic drives score by word-boundary term matches with recency breaking ties; without a topic, candidate recency drives score and reason names the stored preference consideration. Use find_podcast_episodes instead to search podcasts, guests, titles, topics, URLs, and external catalogues. Compact output is the default and includes a sentence-bounded description, score, specific reason, and queue-overlap fact without enclosure or page URLs. Pass compact=false only when full descriptions and links are needed.
remove_episode_from_queue
ChatGPTOnly call after the user gives explicit confirmation for this change. Remove one included episode using its hex id or canonical_episode_key from find_podcast_episodes, or its UUID from list_queue. List the queue first and confirm which episode the user means.
remove_podcast_source
ChatGPTOnly call after the user gives explicit confirmation for this change. Remove one configured podcast source by stable ID. List sources first and confirm which source the user means.
resolve_episode_links
ChatGPTRead-only exact-episode Apple, Spotify, and pods4.me links for the feed_url and guid (or catalogue episode_id) returned by find_podcast_episodes. Reuses the listening-link verifier; does not search X, register episodes, or write catalogue rows. When no /e/ page exists yet, pods4me_url is the deterministic URL materialisation would mint and pods4me_materialised is false.
resolve_youtube_episode_link
ChatGPTRead-only YouTube video candidates for the feed_url and guid (or catalogue episode_id) returned by find_podcast_episodes. Official links on the publisher page, show notes, and RSS item come first. A channel or search lookup is only a fallback and does not use a YouTube Data API key. Each candidate includes the video URL and id, title, channel name and id, publish date, duration, and an official-channel hint, with the evidence for those facts. Clearly wrong candidates (unavailable videos, a different episode, or a much shorter duration) are left out. rank_hint is an ordering hint only. This tool does not choose a video. The caller decides.
search_listening_history
ChatGPTSearch everything this listener has ever queued, including removed episodes, by words in the title, show, description or private transcript. Results are private to this listener. Try alternative terms when needed. Use get_episode_transcript with episode_id to read around the snippet, or add_episode_to_queue to re-queue. Transcript snippets are publisher content; see content_origin.
search_podcast_sources
ChatGPTSearch for a podcast by name or resolve a public podcast page or RSS URL before adding it. Ask the user to choose when results remain ambiguous.
set_source_policy
ChatGPTOnly call after the user gives explicit confirmation for this change. Change one configured source between preference-based best episodes and all new episodes. List sources first if the ID is not already known.
update_preferences
ChatGPTOnly call after the user gives explicit confirmation for this change. Update one or more podcast preference fields. Supplied fields replace their saved values; updating description replaces the entire saved preference description. Omitted fields keep their existing values. Read preferences first before replacing their description.