Serve SDK
Creates a clean Markdown representation of a page and serves it to recognized agents from the site's original URL, while regular visitors continue to receive the full page.
Onto
Onto helps website owners and developers make public pages readable to AI agents. Its Serve SDK delivers Markdown at the original URL, while the Read API, MCP server, scanner, AIO score, and analytics support agent reading and site audits.
AI-readable website delivery and inspection
Onto is a developer tool for making websites easier for AI agents to read. It provides clean Markdown at the site's existing URL through the Serve SDK, and offers a Read API for fetching public pages from application code or an agent workflow.
The product combines delivery, inspection, and measurement: the SDK prepares Markdown at build time, the API can read and score URLs, and the AIO Scanner identifies named readability faults on a public page. An MCP server exposes Onto's reading tools to compatible agents, while the Serve dashboard shows visits from agents that the SDK answers.
Core features
Read API
Fetches public URLs and returns cleaned Markdown with metadata and raw-versus-cleaned size statistics. The API can return JSON or raw Markdown depending on the request.
AIO score
Scores AI readability from 0 to 100 and identifies named structural issues, including crawler access, Markdown availability, schema markup, headings, and image alternative text.
MCP server
Provides six reading tools through Onto's connector or a local npx server, including URL reading, so compatible agents can retrieve page content through MCP.
AIO Scanner and reports
Checks a public page without signup and shows the score, named faults, and the exact Markdown view produced for an agent. Full reports can be saved to an account.
Agent analytics
Records agent visits for sites using the Serve SDK, including the agent, page, location, and payload size. Serve Pro adds unknown-agent reporting and CSV export.
Practical use cases
Improve a public website's agent-readable layer
Install the Next.js SDK so AI crawlers receive a build-generated Markdown version of documentation, pricing, or product pages while human visitors keep the normal rendered experience.
Build web-reading workflows
Use the Read API from an agent, internal tool, or application that needs to retrieve public web pages as structured Markdown instead of processing full HTML.
Audit AI readability
Run the scanner on a page to find access, structure, Markdown, and metadata issues before publishing changes, then use the report to focus remediation on named faults.
Give agents a page-reading tool
Connect an MCP-compatible assistant to Onto so it can call reading tools when it needs page content, with usage charged through the Read account.
Monitor agent traffic and payloads
Review which named and unknown agents visit SDK-served pages, which routes they request, and how much smaller the Markdown response is than the source page.
Pros and Cons
Pros
- Delivers Markdown from the same URL, separating the agent response from the full browser page without requiring a separate public content URL.
- Provides an auditable AIO score with named penalties rather than an unexplained overall rating.
- Offers several access paths: a free scanner, a metered Read API, an MCP server, and an npm SDK for site owners.
- Includes operational safeguards in the Read workflow, such as robots.txt checks, public-host validation, caching, and credit refunds for failed reads.
Cons
- The Read API does not run JavaScript when fetching a page and applies limits such as a 15-second fetch window, up to five redirects, and a 10 MB maximum.
- The Serve SDK documentation cited here is for Next.js 14+ with the App Router, so support for other frameworks is not established by the provided sources.
- The roadmap states that Act, the planned capability for agents to use a site by intent rather than only read it, is not built yet and has no date.
FAQ
How can a team use Onto?
Onto provides three related workflows: the Serve SDK can deliver a Markdown version of prerendered pages from the same URL, the Read API can fetch and clean public URLs, and an MCP server lets compatible agents call Onto's reading tools. The scanner is available without signup for a single-page check.
Is the Serve SDK free, and what framework does it support?
The Serve SDK is available as the MIT-licensed @ontosdk/next package on npm and is free without a usage cap. It creates Markdown copies at build time and uses middleware to serve them to agents at the original URL; the documented SDK support is Next.js 14+ with the App Router.
What does one Read API request return or cost?
The Read API accepts public URLs and returns cleaned Markdown with metadata and size statistics. It can also return an AIO score, read multiple URLs through batch, map URLs from a sitemap, and extract JSON-LD, OpenGraph, and meta information. One URL read uses one credit; scoring is free.
What does the free AIO Scanner check?
The scanner checks one public page without signup, a card, or an API key. It examines robots.txt, the page as a browser without JavaScript, Markdown responses and related paths, then shows a score, named faults, and the Markdown view. A saved report requires an account.
Are Read and Serve billed together?
Read and Serve are separate subscriptions. Read has a free tier with 1,000 credits per month and paid tiers with additional credits and concurrency. Serve has a free SDK tier and paid dashboard plans; paid Serve features include longer history, unknown-agent reporting, CSV export, per-route context, and support for more sites.
Quick Facts
- Category
- Developer tool
- Primary users
- Website owners, developers, and AI application builders
- Website delivery
- Markdown at the original page URL through the Serve SDK
- API output
- Clean Markdown with metadata and read statistics
- Agent access
- Read API, MCP server, and Serve SDK
- Free entry points
- AIO Scanner without signup; Serve SDK free on npm; Read includes 1,000 credits per month
Alternative a Onto
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
GitBot
GitBot is a local browser-based workspace for creating reusable coding bots powered by Claude Code, Codex, or OpenCode. It helps developers run those agents across repository folders with persistent threads and configurable permissions.
AakarDev AI
AakarDev AI helps teams manage AI provider access, project-level setups, logs, and analytics from one dashboard. It supports BYOK workflows and lists providers including OpenAI, Google Gemini, Anthropic, Groq, Mistral AI, and Perplexity AI.
Trigger.dev chat agent
Trigger.dev chat agent is a durable AI chat backend for developers building stateful conversations that can survive refreshes, crashes, and long-running turns. It connects with the AI SDK `useChat` flow and runs on managed infrastructure with no timeout on a turn.
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
Codex Plugins
Codex Plugins bundle reusable skills, app integrations, and MCP servers into workflows you can install in the Codex app or use from Codex CLI. They help extend Codex with connected-service tasks, reusable instructions, and shared team workflows.