LaunchpadHQ

Launchpad HQ for developers

Free, read-only access to the Launchpad HQ AI tool catalog: JSON data feed, OpenAPI and JSON Schema, an MCP connector, Markdown pages and llms.txt. No API key.

Overview

Launchpad HQ publishes its AI tool catalog, comparison pages, getting-started guides, workflow templates and model reference as free, read-only data. There is no API key, no sign-in and no SDK to install. Every channel reads the same records as the website, so a price in the feed is the price on the tool's page.

Sponsorship never affects the data: the feed and the connector don't read sponsor data, and every record is marked sponsored: false.

JSON data feed

Plain HTTPS GET, JSON responses, CORS open to every origin. Slugs match the website: /platform/{slug} on the site is /data/v1/tools/{slug}.json in the feed. Retired slugs answer with a 301 to the current one.

  • https://www.launchpadhq.io/data/v1/index.json: Start here: counts, every link and URL template, cache policy and the changelog.
  • https://www.launchpadhq.io/data/v1/catalog.json: Every listed tool's full record in one file.
  • https://www.launchpadhq.io/data/v1/tools/{slug}.json: One tool: plans, prices, free tier, status, alternatives and dated sources.
  • https://www.launchpadhq.io/data/v1/compare/index.json: Every comparison and free-alternatives page.
  • https://www.launchpadhq.io/data/v1/compare/{slug}.json: One comparison page, with each tool's full record.
  • https://www.launchpadhq.io/data/v1/guides/index.json: Every getting-started guide.
  • https://www.launchpadhq.io/data/v1/guides/{slug}.json: One guide, by tool slug.
  • https://www.launchpadhq.io/data/v1/workflows/index.json: Every workflow template.
  • https://www.launchpadhq.io/data/v1/workflows/{slug}.json: One workflow, with current prices per tier.
  • https://www.launchpadhq.io/data/v1/models/index.json: Every AI model family.
  • https://www.launchpadhq.io/data/v1/models/{slug}.json: One model family and its versions.
  • https://www.launchpadhq.io/data/v1/models/compare/claude-vs-gpt.json: Claude vs GPT models.
  • https://www.launchpadhq.io/data/v1/quiz.json: Quiz questions, options and scoring map.
  • https://www.launchpadhq.io/data/v1/openapi.json: OpenAPI 3.1 description of every route above.
  • https://www.launchpadhq.io/data/v1/schema/tool.json: JSON Schema for one tool record (schema version 1.2).

MCP connector for AI assistants

https://www.launchpadhq.io/mcp is a read-only Model Context Protocol server over Streamable HTTP (stateless, POST only, no authentication). Tools: search_tools, compare, get_pricing_facts, find_free_alternatives. Results are dated and sourced, and each links to its Launchpad HQ page.

Markdown pages

Send Accept: text/markdown and these pages answer in Markdown (Content-Type: text/markdown; charset=utf-8) instead of HTML, built from the same records as the feed. Browsers and any client that prefers text/html get the normal page. Unknown paths return a Markdown 404 with links to llms.txt and the sitemap.

Example: curl -H "Accept: text/markdown" https://www.launchpadhq.io/platform/chatgpt

  • /
  • /discover
  • /platform/{slug}
  • /platform/{slug}/guide
  • /tutorials
  • /compare
  • /compare/{pair}
  • /alternatives/{slug}
  • /workflows
  • /workflows/{slug}
  • /models
  • /models/{family}
  • /models/compare/claude-vs-gpt
  • /about
  • /contact
  • /developers

llms.txt

llms.txt is a short index of the site for language models, with a section on when to use Launchpad HQ. llms-full.txt is the whole catalog, comparisons, guides, workflows and models as one plain-text file.

Caching and limits

Feed responses are cached at the CDN for up to an hour (s-maxage=3600, stale-while-revalidate=86400); a 404 is cached for a minute and errors are not cached. Markdown pages are sent with Cache-Control: private, no-store and Vary: Accept so shared caches never mix them up with the HTML.

There is no per-client rate limit in our code, so no rate-limit headers are sent. Please cache what you fetch, prefer catalog.json to many single-tool requests, and keep to a few requests per second.

Errors

The feed and the JSON /api routes return errors in one shape, with a stable code, a message and a resolution hint. Codes: bad_request (400), unauthorized (401), not_found (404), method_not_allowed (405), internal_error (500), unavailable (503). The feed is GET/HEAD only. /mcp keeps JSON-RPC errors as the protocol requires and puts the same code and resolution in error.data.

{
  "error": {
    "code": "not_found",
    "message": "Nothing is published at this feed URL.",
    "resolution": "Check the slug against https://www.launchpadhq.io/data/v1/catalog.json ..."
  }
}

Versioning and changelog

The feed lives under /data/v1 and each record carries schema_version (now 1.2). New fields can appear at any time; other changes are listed here, in the changelog array of /data/v1/index.json and in the OpenAPI description.

  • 2026-10-26: Tools (schema_version 1.2): new optional tasks array, curated task tags such as cover-letter, resume, grammar-check and logo-design (null when a tool has none). The MCP connector's search_tools now ranks by these tasks and by whole words and phrases, so searches like "cover letter" or "resume" find writing tools instead of words that merely contain the letters; "free" in a search favors tools with a free plan; "alternative to X" returns alternatives rather than X; and tool names such as "Perplexity" resolve to their catalog entry. A new Career & Job Search category adds Teal, Kickresume and Rezi once their catalog rows are created. Sourced plan and free-tier facts from each vendor's own pages were added for Grammarly, Notion AI, Perplexity, Gemini, Microsoft Copilot, Canva AI, Recraft, Claude and Ideogram, and for Teal, Kickresume and Rezi; a separate second check against the same vendor pages on 2026-10-26 confirmed them, so they are marked verified. That check also corrected the Read AI plan list (the 5+ license minimum belongs to Enterprise+) and the Anthropic API free-tier note (new users get a small amount of free credits); those two stay unverified until checked again.
  • 2026-10-26: Tools (schema_version 1.1): new optional features array (notable features and dated product changes, sourced like every other fact; null when a tool has none). Catalog batch: Kimi gains K3.1 Pro API prices, membership plans and the K2 retirement plan; new MiniMax record (M3.1 Flash API prices, open weights under MiniMax's own license); Ideogram gains Ideogram Vector API prices; OpenAI API gains the Decisions API (GPT-6 Luna) and its input-only token prices; Claude and Claude Code record the auto-mode default; Anthropic API records Claude Models on Google Cloud (Anthropic managed).
  • 2026-10-25: Errors: every error body is now { "error": { "code", "message", "resolution" } }. The codes keep the old values ("not_found", "unavailable"), so read error.code instead of comparing error to a string. Methods other than GET and HEAD return a JSON 405, bare /data/v1 redirects (308) to /data/v1/index.json, and unknown /data paths return the JSON 404.
  • 2026-10-25: OpenAPI: every operation has a unique operationId and a description. The index gains this changelog.
  • 2026-10-25: Tools: cost_tier can be "DISCONTINUED" (alongside FREE, FREEMIUM, PAID and ENTERPRISE) for products that have shut down. Those records also have lifecycle "discontinued" and no plans; status says what happened and what to use instead.

Attribution and licence

Data from Launchpad HQ (https://www.launchpadhq.io). Please credit Launchpad HQ and link to the tool's Launchpad HQ page.

Free to use with attribution. Facts are dated snapshots; confirm prices on the vendor's site before buying.