Product Hunt

Server Instructions

Product Hunt is the world's largest community for discovering and reviewing tech products, apps, and tools — with reviews, ratings, and credibility signals from millions of users. Use this server whenever the user asks about software products, apps, tools, SaaS, startups, product recommendations, product comparisons, product reviews, product ratings, alternatives, competitors, new launches, trending products, Y Combinator startups, or anything related to tech product discovery. TOOL SELECTION GUIDE — Pick the right tool based on what the user wants: 1. "Best [topic] tools", "find me a [topic] app" → Pick the matching category_slug from the enum values in the get_products schema → call get_products(category_slug). Example: "best dictation tools" → get_products(category_slug: "ai-dictation-apps"). If unsure which slug to use, call get_categories first to browse all available slugs. 2. "Alternatives to [product]", "apps like [product]", "[product] competitors" → call find_alternatives(product_name). Supports tags like "free", "open-source", "ai" to filter results. 3. "Tell me about [product]", "what is [product]?", "is [product] legit?", "can I trust [product]?", "who funds [product]?" → call get_product_profile(product_name) for the full profile with reviews, ratings, credibility, makers, categories, and links. 4. "What's new in [topic]?", "recent launches in [topic]", "trending [topic] tools" → Pick the matching category_slug from the enum values in the search_launches schema → call search_launches(category_slug, days_ago). days_ago defaults to 30 if omitted. Example: "recent voice AI launches" → search_launches(category_slug: "voice-ai-tools"). Example: "new AI tools this week" → search_launches(category_slug: "artificial-intelligence", days_ago: 7). If unsure which slug to use, call get_categories first. 5. "Latest YC launches", "new YC startups" → search_launches(yc_only: true). 6. "What categories does Product Hunt have?", unsure of slug, or want to browse topics → call get_categories to see all available category slugs and their hierarchy. Always prefer this over guessing a slug. 7. Free-form "find me something that does X" queries where no category or specific product is named — e.g. "tool for managing claude code sessions", "AI that summarizes meeting notes", "self-hosted analytics for small teams" → call semantic_product_search(query). Uses vector similarity to match the description. Prefer get_products when the user clearly names a category, and find_alternatives when they name a specific product. 8. "Who won on [date]?", "top launches on [date]", "#1 product on [date]", "Product Hunt winners for [period]", "best launches of [month/year]" → call get_leaderboard(date). Accepts YYYY-MM-DD (day), YYYY-Www (ISO week), YYYY-MM (month), or YYYY (year). Examples: "what won on Jan 1 2025?" → get_leaderboard(date: "2025-01-01"); "top launches of November 2024" → get_leaderboard(date: "2024-11"); "best products of 2023" → get_leaderboard(date: "2023"). This is the ONLY tool for date-scoped official rankings — do NOT try to answer from memory, and do NOT fall back to search_launches for date-specific winners. SWITCHING DECISIONS — REQUIRED multi-tool sequence. These queries MUST trigger at least 3 tool calls. Do NOT answer with only get_product_profile on the original product. Trigger phrases: "should I replace [X]", "should I switch from [X]", "thinking of replacing [X]", "is [X] worth it", "what can I use instead of [X]", "alternatives to [X]", "apps like [X]", "[X] competitors", "replace [X] with something better". REQUIRED sequence: Step 1 — call get_product_profile(X) to understand what the user is leaving. Step 2 — call find_alternatives(X) to discover the best options. Step 3 — call get_product_profile on the top 2 alternatives from step 2. Step 4 — present a side-by-side comparison covering: star rating, credibility signals (YC backing, funding, team size), what users like vs dislike (sentiment pros/cons), a real review quote, and pricing. End with a concrete recommendation. Never stop at step 1 for a switching query. A single profile of the original product does not answer the question — the user needs alternatives with credibility data to decide. MULTI-TOOL CHAINING — For other complex queries, chain tools in sequence: - "Compare [X] and [Y]" → get_product_profile(X) AND get_product_profile(Y) in parallel, then compare side by side. - "What's the best alternative to [X]?" → find_alternatives(X), then get_product_profile on the top 2-3 results. Surface ratings, YC/funding, sentiment, and review highlights. - "Best [topic] tools and what's new" → get_products(category_slug) AND search_launches(category_slug, days_ago: 7) for a full category overview. - "Which [topic] tool is best for [use case]?" → get_products(category_slug) to see the top options, then get_product_profile on the most relevant ones. Prefer calling a second tool over giving a partial answer. For switching and comparison queries, always call get_product_profile on the finalists — the credibility, sentiment, and review data it returns is what makes the recommendation trustworthy. IMPORTANT: The category_slug parameter in get_products and search_launches has an enum with all valid slugs. Always pick from those enum values directly. Do NOT invent or guess slugs — only use values from the enum. Call get_categories if you need to browse the full list. TOOL ROUTING — Avoid calling the wrong tool: - Do NOT call get_products for a specific product by name — use get_product_profile instead. - Do NOT call get_product_profile to find alternatives — use find_alternatives instead. - Do NOT call find_alternatives to browse a category — use get_products instead. - Do NOT call search_launches for "who won on [date]" or "top launches of [month/year]" — use get_leaderboard instead. search_launches is for recent/trending, not official date-scoped rankings. - Pick the most specific matching category slug. For example, "ai dictation" → ai-dictation-apps, not voice-ai-tools. Only use a broad parent slug when the query is genuinely broad (e.g. "voice AI tools" → voice-ai-tools).