Microlink turns any link into screenshots, PDFs, previews, or structured data. No more Puppeteer clusters. Open source roots. Say hi to [email protected].
Joined September 2017
- Tweets279
- Following0
- Followers588
- Likes7
Pinned Tweet
unavatar.io is now generally available
- One endpoint for avatar resolution
- Usernames, emails, and domains
- Free tier keeps 50 avatars per day
- Pay-as-you-go at $0.001 per token
- Only charged for successful, uncached avatars
microlink.io/blog/unavatar-g…
You can make your AI agents understand Microlink in under a second.
Just run `npx microlink.io setup`
OpenAI agents used Microlink to escape their sandbox.
The Microlink API let them bypass sandbox restrictions, turning a product feature into an escape hatch.
Full story:
microlink.io/blog/how-microl…
Sometimes a product's biggest strength is its biggest marketing challenge.
Our API at Microlink spins up native browsers. If a human can do it in a browser, our API can automate it.
We provide the raw engine. You decide where to drive it:
microlink.io/api
Introducing a new free tool: get every URL from a website’s sitemap.
microlink.io/tools/sitemap
Useful for crawling, SEO audits, content discovery, research, or simply figuring out what a website actually exposes.
Paste a sitemap URL, get the URLs.
Programmatic web fetching shouldn’t mean fighting anti-bot walls.
We optimized our Smart Proxy layer:
- Triggers only when needed to keep latency low
- Escalates proxy strength as restrictions increase
- Supports custom proxy.location
microlink.io/features/proxy
We increased the memory limits for Microlink Functions:
- Free: 64 MB
- Pro: 128 MB
Also, a new playground:
microlink.io/function
Turns out Microlink has been helping OpenAI agents cheat.
They were sandboxed to GET-only. So they sent us a GET that spun up a browser on our servers and ran the POST they weren't allowed to make.
Happy to help, I guess.
microlink.io/blog/how-microl…
You can now use Microlink CLI from the web:
microlink.io/terminal
Interact with Microlink on the go, wherever you are.
Our antibot system now retries requests that require a specific origin location.
You can also set it explicitly using the proxy.location parameter.
180+ locations supported: microlink.io/docs/api/parame…
Shipped microlink SDK.
Pass a link, a real browser opens it, take what you need.
◆ 20 methods: screenshot, pdf, markdown, links, emails…
◆ antibot built in, ideal for web-browsing agents
◆ isolated browser per call, parallel as needed
microlink.io/integrations/sd…
Handinger came to us because their agents needed live Google (search, maps, news) without maintaining a scraper.
They build AI agents from plain-English briefs: prospect research, competitor monitoring & lead enrichment.
Every web task runs on our api.
microlink.io/use-cases/handi…
microlink.io retweeted
Totally obsessed with this screenshot API.
I've integrated it into three products so far.
microlink.io/screenshot
microlink.io retweeted
The numbers behind @microlinkhq for the last 6 months:
→ 2.3B requests served from our edge. 57TB of traffic.
→ 12.7M reqs/d or 8,800 reqs/min or 147 reqs/s
→ 345M screenshots generated. ~1,300 every minute, 24/7, for six months.
→ 300M+ pages fetched. 32M rendered in real Chrome.
→ 99.92% availability. Slightly above our SLA.
→ 640K+ anti-bot challenges cleared across 30 vendors: Cloudflare, reCAPTCHA, Turnstile, DataDome, Imperva, Akamai, PerimeterX, hCaptcha, and more.
→ Busiest day: April 7. 8.7M responses. Peak: 160 req/s.
→ 220 browser pages open simultaneously at peak.
Just me and @Josebaseba.
Small team. Big infrastructure.
Microlink now supports retrieve logos from BIMI recods.
• Official SVG brand logos
• DNS-first lookup
• Fallback to existing strategies
microlink.io/blog/brand-logo…
We rebuilt the homepage around a live playground.
Pick an action in the hero and it fires the same request you'd write yourself, so what you get back is the real response — URL, headers, timing, and the code.
Take a look!
microlink.io
microlink.io retweeted
Microlink Markdown capabilities keeps getting better.
You can now turn any YouTube URL into clean markdown.
Point it at a video, get the transcript back as markdown. No scraping, no yt-dlp glue. Ready to feed an LLM.
