One API call before every action. Know what's allowed. Stop guessing.
Agents waste tokens, time, and trust discovering at runtime what could be resolved in a single lookup. Maango is the registry that tells your agent what every domain allows, before it acts.
Free tier. No credit card. 1,000,000+ domains in registry.
A federal judge blocked Perplexity's Comet agent from Amazon, ruling user permission ≠ platform permission.
Read the rulingSites blocking AI agents jumped 336% year over year. ClaudeBot now blocked at 5.8M+ domains. Your agent is failing more often than ever.
See the dataArticle 53 transparency obligations hit any agent operating in or serving EU users in 2026. The audit clock is ticking.
A blocked attempt isn't a 403. It's the agent reasoning over the 403, planning a retry, picking an alternative, executing again. Thousands of tokens per failed action, multiplied across every user.
Discovery at runtime takes seconds. Discovery at lookup takes milliseconds. Every blocked attempt is latency your user feels.
When legal asks what your agent did and where it went, "we ran an LLM" is not an answer. Compliance teams need a per-action policy decision log.
# Before any web action - one GET against the Maango registry. policy = requests.get(f"https://api.maango.io/v1/domain/{domain}", headers={"Authorization": f"Bearer {API_KEY}"}).json() if policy["use_cases"]["inference"] == "blocked": agent.skip(url, reason=policy["stance"]) else: agent.fetch(url)
Drop into LangChain, Browser Use, Stagehand, custom loops, or anything that hits a URL. Resolves in under 100ms. Free tier covers most builds.
One call returns a structured decision per domain across training, search, and inference. 8 standards aggregated into a single answer. No more parsing robots.txt by hand.
Every policy check is logged with timestamp, domain, and decision. Pull your full decision history when compliance asks. Built for legal, by-design.
robots.txt, llms.txt, ai.txt, TDMRep, Content-Signal, Markdown for Agents, AI meta tags, and ToS clauses. We resolve conflicts so your agent gets one clean answer.
200 requests/min. 10K/day. 100K/month. No credit card. The API is in beta and we're shipping fast - design partners help shape what's next.
Don't repeat Comet's mistake. Pre-flight checks tell you which retailers permit agentic transactions and which will block, sue, or both. Route to the yeses, log the rest.
Public-web doesn't mean permission-given. Maango tells your agent which domains have published policies, which are silent, and which actively block, so you stay defensible while still moving fast.
Agents acting on behalf of employees inside enterprises need a defensible audit posture. Every web action runs through Maango first. Every decision is logged. Your customer's compliance team has receipts.
If you're selling agents to enterprise buyers, the question isn't whether your agent works. It's whether the legal team will let you deploy. Maango gives every web action a logged, structured policy decision your buyer's compliance review can defend. We work directly with design partners building for regulated buyers.
Talk to us about design partnershipAPI Response Preview
Today, Maango reads every AI policy standard because no single one is authoritative. robots.txt wasn't built for AI. llms.txt is optional. Content-Signal only covers Cloudflare's slice. ai.txt is Spawning's optional spec. The web has 8 competing formats and no source of truth.
That's the gap we're closing. As website owners declare policies directly through Maango, the registry becomes the canonical layer, not a scraper of fragments, but the place where AI permissions live. One registry, every agent checks it. Like DNS, but for AI permissions.
Run a website? Set your AI policy in 2 minutes and let every Maango-checking agent respect it. Go to Maango for publishers
90% of the web has no AI policy. Whether you're building agents or running a website, you need a source of truth.