For developers and AI agents
Read Aback Launch from code or from an AI agent: an MCP server and a REST API, both free, read-only and without keys.
MCP server
Endpoint: https://abacklaunch.com/mcp. Streamable HTTP, stateless: POST JSON-RPC and get a JSON answer. Add it to any MCP client as a remote server; no authentication is needed. To let an agent read your own listings and stats, connect your account when the client asks (OAuth, read-only), and disconnect it any time in Settings.
search_aback_launchSearch launched products, guides, categories, tags and pages on Aback Launch.get_top_launchesThe top products launched on Aback Launch this week, this month or of all time, in leaderboard order.get_launchOne product launched on Aback Launch: what it does, who it is for, pricing and platforms, its FAQ and links.find_launch_platformsWhere to launch or list a product: launch platforms, startup and AI directories, review sites and communities, with who each suits and whether its link is dofollow (only when checked on real listings).get_launch_platformOne launch platform or directory in full, with a step-by-step launch guide when there is one.compare_launch_platformsTwo launch platforms side by side, with a written verdict on which to choose when there is one.check_character_limitCount text the way a platform field does (Product Hunt tagline, App Store subtitle, Google Play short description, X post...) and say whether it fits and what gets cut off.define_termA glossary definition (dofollow link, GEO, Show HN, launch week...). Without a slug, lists every term.list_promptsThe prompt library: task briefs a coding agent can follow to fix a site's SEO, AI visibility and launch readiness (pre-launch audit, llms.txt, structured data, sitemaps...). Each has a slug for get_prompt.get_promptOne prompt from the library, rendered for the site's stack: the job, ground rules (read first, plan, wait for approval), limits, steps and what done looks like. Follow it in the user's codebase, starting with the plan.list_free_toolsThe free tools on Aback Launch: SEO and AI search checkers, launch helpers, character limit and image size checkers.askA question in plain words ("free launch platforms with dofollow links", "a prompt to add llms.txt"), answered only from Aback Launch's own pages: listings, the launch platform database, prompts, tools and the glossary. Returns the matching pages with links, and a short cited answer when answers are turned on.get_my_accountThe signed-in member's own listings on Aback Launch: review status, launch date, votes, AI-readiness score and badge, with page views, visitors and clicks for the last 30 days. Needs the member to connect their account (OAuth, scope account:read).
REST API
Base URL: https://abacklaunch.com/api/v1. OpenAPI 3.1: /api/v1/openapi.json. Responses are JSON, cached at the edge, and open to any origin (CORS).
GET /meGET /launchesGET /launches/{slug}GET /searchGET /launch-platformsGET /launch-platforms/{slug}GET /launch-platforms/compareGET /character-limitGET /glossaryGET /glossary/{slug}GET /promptsGET /prompts/{slug}GET /tools
Example: https://abacklaunch.com/api/v1/launches?window=weekly&limit=5
Limits and rules
- 120 requests a minute per IP across the API and MCP server; search allows 30. Over the limit you get HTTP 429 with Retry-After.
- Launches appear once they are live, with the fields their public page shows. Nothing private is ever returned.
- Link types are only stated when read from real listings; “not-checked” means unknown.
- Submitting, voting and paying need a signed-in person on the website; there is no write API.
- Link to the launch page when you show a launch. Health: /api/health.
Agent skills
- aback-launch-data: Read Aback Launch data: top and recent launches, site search, the launch platform database with checked link types, platform comparisons, character limits and glossary terms, through the MCP server or the REST API.
- launch-on-aback-launch: Help a founder launch a product on Aback Launch: what to prepare, how the weekly Friday launch works, how the free review and the dofollow badge work, and what needs the founder themselves.
- choose-launch-platforms: Choose where to launch a product using the Aback Launch platform database: match platforms to the product, compare them, check which give a dofollow link, and plan the launch order.
- pre-launch-audit: Audit the whole site before launch and get a ranked fix list, then fix the blockers.
- launch-day-traffic-readiness: Make the app survive a sudden burst of visitors: caching, limits, errors and a rollback plan.
- launch-analytics-attribution: Know which launch site, post and AI assistant sent each signup, not just how many visitors came.
- social-share-cards: Make every link to your site preview with the right title, description and image, on every network.
- waitlist-launch-email: Email your waitlist on launch day without landing in spam: clean list, authenticated domain, batched sends.
- waitlist-landing-page: Build a pre-launch page that says what is coming and collects real, confirmed sign-ups.
- llms-txt-and-markdown: Add llms.txt and a Markdown version of every public page, so AI assistants read your site cleanly.
- ai-crawler-access: Decide which AI bots can read your site, write it in robots.txt, and make sure nothing else silently blocks them.
- answer-ready-content: Give each key page a one-sentence definition, a facts table and real FAQs that assistants can lift as answers.
- structured-data: Add correct, honest JSON-LD for each page type from one helper, matching what the page shows.
- citable-stats-page: Publish your real numbers on a page writers and AI assistants can cite, with dates, method and anchors.
- entity-and-founder-links: Tie your site, company, founder and official profiles together so search and AI know they are one entity.
- titles-descriptions-canonicals: Give every page a unique title, a real description and the right canonical, from one helper.
- sitemap-and-indexnow: Generate the sitemap from your real pages with true dates, and ping IndexNow when content changes.
- redirects-and-404s: One canonical host, no redirect chains, real 404s and 410s, and every old URL sent somewhere useful.
- internal-linking: Find orphan pages and broken links, and add related links from your data so every page is reachable.
- image-seo: Give every image useful alt text, a modern format, the right size and correct lazy loading.
- core-web-vitals: Find what slows your main pages and fix LCP, INP and CLS at their causes.
- programmatic-pages: Make generated pages worth indexing, and keep thin ones out of the index and the sitemap.
- hreflang: Tell search engines which language version of each page to show, with correct, reciprocal hreflang.
- landing-page-above-the-fold: Make the first screen say what the product is, who it is for and what to do next, and load fast.
- pricing-page: Make plans clear, prices match checkout, and the common pricing questions answered on the page.
- first-run-onboarding: Shorten the path from sign-up to the first moment a new user gets value, and measure it.
- empty-and-loading-states: Give every list and screen a helpful empty state, a calm loading state and an error with a way out.
- signup-friction: Cut sign-up to the fewest fields, add a faster sign-in option, and stop it blocking the first visit.
- testimonial-wall: Show real quotes from real users, with permission and attribution, from one data file.
- security-headers: Set HSTS, CSP, framing, MIME and referrer headers on every response, with CSP rolled out without breaking the site.
- auth-hardening: Check how passwords, sessions and resets work, and fix the gaps attackers try first.
- secrets-scan: Find API keys and passwords in the code, the history and the browser bundle, and stop new ones getting in.
- rate-limits: Limit how often anyone can hit the endpoints that cost money or invite abuse, with clear 429 answers.
- error-tracking-uptime: Know about errors and outages before your users tell you: error tracking, a health endpoint and alerts.
- database-backups: Make sure your data is backed up on a schedule, kept long enough, and that a restore actually works.
- email-authentication: Find every service that sends as your domain and get the exact SPF, DKIM and DMARC records to add.
- transactional-emails: Make every email your app sends consistent, readable in any client, and easy to reply to.
- unsubscribe-and-consent: Record consent, add one-click unsubscribe, and make sure nobody who opted out is emailed again.
- privacy-and-cookies: List what the code really collects and shares, then make the policy and the cookie banner match it.
- data-export-deletion: Let users download their data and delete their account themselves, covering every table and service.
- accessibility-pass: Make your main pages and flows usable by keyboard and screen reader, with readable contrast.
- changelog-page: Build a changelog from structured entries, with permalinks, a feed and a link from inside the app.
- press-kit-page: Build a press kit page with a boilerplate, facts, founders, downloadable logos and screenshots, and a contact.
- featured-on-section: Show the launch badges and mentions you earned in one tidy, fast section, from one data file.
- comments-to-faq: Group the questions people asked on launch day and turn them into an FAQ on the pages where they come up.
- feedback-triage: Turn launch feedback into reproduced bugs, grouped requests and a ranked list of quick wins.
- relaunch-changelog: Work out what changed since the first launch and turn it into a changelog entry and a relaunch post draft.
- exit-survey: Ask people who cancel one optional question about why, store the answers, and never make cancelling harder.
Discovery files
- /.well-known/ai-catalog.jsonCapability manifest (ARD) listing everything below
- /.well-known/api-catalogAPI catalog (RFC 9727)
- /openapi.jsonOpenAPI 3.1 description of the REST API
- /.well-known/mcp/server-card.jsonMCP server card
- /.well-known/agent-skills/index.jsonAgent skills index
- /auth.mdHow agents register and sign in (optional; read-only)
- /.well-known/oauth-authorization-serverOAuth authorization server metadata (RFC 8414)
- /.well-known/oauth-protected-resourceOAuth protected resource metadata (RFC 9728)
- /llms.txtA map of the site for language models
- /robots.txtCrawl rules and Content Signals
- /.well-known/http-message-signatures-directoryThe key our crawlers sign their requests with (Web Bot Auth)
Every page also answers as Markdown: send Accept: text/markdown or add .md to the URL. Questions: contact us.