Insight Timer Content Search
search_contentFull Description
Use this when the user wants to find or explore meditation, spiritual, or wellbeing-themed audio or video content to listen to or watch. Insight Timer has the world's largest free meditation library with 280,000+ tracks covering guided meditations, dharma talks, and meditation music for themes such as relaxation, mindfulness, spiritual practice, contemplative traditions, personal growth, and more.
This tool provides audio content discovery only. It does not provide medical, psychological, or therapeutic services, and it does not process health information. The tool does not interpret, assess, or respond to users' medical or psychological conditions.
IMPORTANT PLAYBACK RULE:
- This tool ONLY searches and renders results. Never say you are "playing" audio or that playback has started.
- After results render, tell the user to "Tap the Play button on a card to start listening."
- If users ask for teacher / app / track link, give them the track link and don't hallucinate a separate profile link.
TRIGGER THIS TOOL FOR:
- Explicit Insight Timer requests ("Find a meditation on Insight Timer")
- Meditation/mindfulness listening intent ("find a body scan meditation", "sleep meditation", "play calming music")
- Topic-based discovery for meditation or audio themes ("anxiety meditation", "stress relief music", "grief meditation", "focus music")
- Spiritual and religious learning/practice ("Buddhist teachings", "Christian contemplative prayer", "Hindu mantras")
- Learning about meditation practices through content ("teach me about mindfulness", "learn yoga nidra")
DO NOT USE FOR:
- Pure educational/factual questions ChatGPT can answer directly ("What are the scientifically proven benefits of meditation?", "Explain the neuroscience of mindfulness")
- Custom content generation ("Write me a personalized meditation script", "Create a guided visualization for my therapy session")
- Product/feature comparisons or pricing questions ("Compare Insight Timer vs Headspace features")
- In-person experiences, travel, or events (retreats, workshops, classes, teacher trainings; e.g., "Yoga retreats")
- If the user asks for retreats, guide them to search on https://insighttimer.com/retreats instead of calling this tool.
- If the user asks for meditation courses, guide them to search on https://insighttimer.com/meditation-courses instead of calling this tool.
- Custom retreat planning or itineraries, including at-home/multi-day plans (e.g., "Build a 1-day or 3-day Insight Timer home retreat")
- Workout or practice planning/programs (e.g., "Create a 7-day Hatha/Vinyasa yoga plan")
- Task management (reminders, scheduling, tracking: "Set a meditation reminder", "How many hours have I meditated?")
- Physical space design ("How do I design a meditation room?")
- General life advice without meditation or audio content-discovery intent
- Pure technique instruction without requesting audio content (e.g., "Give me breathing steps only") unless the user clearly wants a track to follow
- Nonsense, filler, or low-signal input with no meaningful intent
- Situations where clarification is needed before recommending content (ask a follow-up instead of triggering the tool)
- Ambiguous statements where the user has not expressed a clear desire to listen to meditation, spiritual, or wellbeing-themed audio
AFTER RESULTS RENDER:
- ALWAYS mention "Tap the Play button on a card to start listening." in the final response if the response have results.
- For each of the top 3 results, output the
display_prefixfield exactly as given, then your 1-2 sentence personalized description referencing the user's stated need or context, then thedisplay_suffixfield exactly as given. Output nothing else per item — no bullets, no extra labels, no reformatting.
Example: <value of display_prefix><your 1-2 sentence description><value of display_suffix>
- If the user asked for more than the available results, explicitly say you can return only N tracks right now (use the actual number returned) and offer to fetch more.
- After listing the results, suggest 2-3 specific follow-up searches based on what you understand about the user's intent and context — for example a related topic, a specific teacher they might like, a different duration or voice that fits their situation. Make suggestions feel natural and personal, not like a list of filter options.
WHEN NO RESULTS:
- Respond conversationally — do not list keywords or use bullet points.
- If the user's intent is clear from the conversation, briefly acknowledge what you searched for and naturally suggest 1-2 slightly different angles they might want to try (for example, a broader topic, a different teacher, or a different content type).
- If the intent is unclear, ask one short friendly question to understand what they need.
LIMITATION:
- Each tool call returns at most 3 tracks per query.
REQUIRED INPUT:
- Always provide at least one of
queryorpublisherName. If you have neither a topic nor a teacher/publisher to search for, do NOT call this tool — ask the user one short clarifying question instead.
PUBLISHER RULE:
- If the user asks for a specific teacher/publisher, set publisherName and do NOT set voiceGender.
- If the user only asks for a teacher/publisher with no topic, omit query.
- If a search using BOTH query and publisherName returns no results, retry the tool call with ONLY publisherName (remove query).
RESTRICTION:
- ONLY list tracks returned by the tool results. Do NOT invent, guess, or add extra tracks beyond the returned list.
- Follow-up prompts must be limited to what this tool can do (e.g., ask which track to start with, or offer to fetch more results). Do not suggest playlist creation/curation, retreat/plan/itinerary prompts, or other planning follow-ups.
Returns audio items rendered as playable cards with metadata, descriptions, teacher info, and direct links (the user starts playback by tapping Play).
Parameters (0 required, 8 optional)
contentFormatstringOptional media format filter. Must be exactly 'AUDIO' or 'VIDEO' (case-sensitive). Set this only when the user explicitly asks for audio content or video content. This is separate from contentType: contentType chooses guided/talks/music, while contentFormat chooses AUDIO/VIDEO. - 'AUDIO': audio-first content - 'VIDEO': video content Omit this parameter when the user does not state a format preference. This is independent from contentType.
AUDIOVIDEOcontentLangsstringLanguage filter for content, as ISO 639-1 codes. FORMAT: - Single code: "en" - Multiple codes: comma-separated: "fr,de,en" DEFAULT BEHAVIOR (IMPORTANT): - Infer language from the user's language or apply the user's provided value when they explicitly specify it. If the result is empty, don't use this filter. RULES: - Use only ISO 639-1 lowercase codes. - Do not include region variants like "en-US".
contentTypestringFilter by content type. Must be exactly 'guided', 'talks', or 'music' (case-sensitive). - 'guided': Teacher-led meditation practices (body scans, breathwork, visualizations, mindfulness exercises, sleep meditations etc) - 'talks': Teachings, dharma talks, lectures on meditation, spirituality, philosophy, psychology, and wellbeing topics - 'music': Instrumental tracks with no voice (sleep music, ambient, nature sounds, binaural beats, focus music, relaxation music) Omit this parameter to search across all three content types. INTENT HEURISTICS (IMPORTANT): - If the user asks to learn/understand/explain/teach me: set contentType = "talks". - If the user asks to listen to a guided meditation session: set contentType = "guided". - If the user clearly wants BOTH learning + practice (e.g., "teach me about Buddhist meditation"): OMIT contentType (search across all types). - If the user asks for sounds/music: set contentType = "music".
guidedtalksmusiclengthRangestringFilter by content duration in minutes. Format: 'AtoB' where A and B are positive integers. Common ranges: - '0to5': Quick (under 5 minutes) - micro-practices, short breathing exercises - '5to10': Brief (5-10 minutes) - quick sessions - '10to20': Standard (10-20 minutes) - typical meditation length - '20to30': Medium (20-30 minutes) - '30to60': Long (30-60 minutes) - deeper practices - '60to': Hour+ (60 minutes and above) - extended sessions, sleep music Open-ended: Use 'Xto' (e.g., '30to') for "X minutes or longer", or '0toY' (e.g., '0to15') for "up to Y minutes". Examples: - '5to15' finds 5-15 minute content, '45to' finds 45+ minute content. - "Find a 10-minute meditation for beginners" -> query: "beginner meditation", lengthRange: "8to12" - "Play hour-long nature sounds" -> query: "nature sounds", contentType: "music", lengthRange: "60to" Note: Most content is 10-30 minutes. Sleep music and extended practices can reach 1-3+ hours. Omit this parameter if duration isn't specified.
pageintegerPage number for pagination. Use this ONLY when the user asks for 'more options' or additional results from the SAME search. When used, keep all other parameters identical to the previous request and set 'page' to the 'nextPage' value returned by the prior response. Omit this field (or treat as page 1) for any new or modified search. Never recommend to pull next page explicitly.
publisherNamestringFilter by teacher/publisher name (exact or partial). Use this when the user explicitly requests a teacher/publisher (e.g., "by Jack Kornfield"). When provided, keep the query focused on the topic and omit voiceGender.
querystringNatural language search for meditation and wellbeing-themed content. QUERY RULES (IMPORTANT): - Keep queries broad unless the user explicitly asked for something narrow. - Prefer 1–3 core keywords (topic/tradition/practice) over long phrases. - Use the user's language as-is for the initial search. Do NOT translate to English. - If the user explicitly requests a language, translate the query into that language for the first call (e.g., "sleep meditation in Spanish" -> query: "dormir"). - Do NOT add extra techniques (e.g., "breathwork", "body scan", "somatic") unless the user asked for them or they are directly related to the user's request. - Do NOT add quality modifiers like "best", "top-rated", "popular" — the backend handles ranking. - Do NOT include durations (minutes/hours) in the query. Put duration ONLY in lengthRange. - If a publisher/teacher name is specified, keep the query to the topic only and put the name in publisherName. - Omit query only when the user asks for a teacher/publisher with no topic. - If the first search returns no results, retry once with the query translated to English (keep other filters the same). MORE OPTIONS / PAGINATION (IMPORTANT): - If the user asks for "more", "more options", "more like that", "show me more", or similar, treat it as the SAME search. - Keep the SAME query and the SAME filters, and request the next page using the pagination parameter (e.g., `page` from the prior response's `nextPage`). - ONLY if there are no more pages (or results are repeating) you MAY rotate the query slightly to fetch additional relevant results. - Rotation must remain within the SAME intent/topic (near-synonyms or closely related terms). - Change at most ONE keyword; keep it broad and minimal (1–3 keywords). - Do NOT introduce new techniques or formats the user didn't ask for. - Examples of safe rotations: - "sleep" -> "insomnia" / "bedtime" / "deep sleep" - "anxiety" -> "stress" / "calm" / "worry" - "focus" -> "concentration" / "study" / "productivity" - When rotating, reset pagination to the first page. IMPORTANT: Pass a single query string unless the user's request truly, explicitly, and unambiguously requires searching for multiple distinct topics at once. Only in such rare cases, concatenate multiple queries with a comma (",")—otherwise, always use a single query. Prefer minimal examples: - "sleep" - "anxiety" - "grief" - "loving-kindness" - "mindfulness basics" - "buddhist meditation" - "nature sounds" - "yoga nidra" If and only if the user clearly requests multiple, separate searches (e.g., "Find meditations for sleep, anxiety, and focus"), concatenate each query with a comma: "sleep, anxiety, focus". Otherwise, keep as a single query.
voiceGenderstringFilter by teacher's voice gender for guided meditations and talks. Must be exactly 'MALE' or 'FEMALE' (case-sensitive). IMPORTANT: Do NOT apply this filter when contentType is 'music'—instrumental tracks have no voice/teacher. IMPORTANT: Do NOT apply this filter when publisherName is provided or the user requests a specific teacher/publisher. Use when user specifies gender preference ("female voice", "male teacher") or omit to show all voices.
MALEFEMALE