Onto MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ONTO_API_KEY | Yes | Your Onto API key from app.buildonto.dev/read/keys | |
| ONTO_API_BASE | No | Override the API base URL (default: https://api.buildonto.dev) | https://api.buildonto.dev |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| read_urlA | Read any URL and return clean, agent-ready Markdown. Strips HTML noise, preserves semantic content, and returns content optimized for AI consumption. Use this when you need to extract content from a website for an AI agent to process. |
| score_urlA | Get the AIO (AI-readability) score for any URL. Returns a 0-100 score plus a list of penalties, benefits, and recommendations describing why the source is or is not well-suited for AI consumption. Use this to evaluate source quality before relying on it. |
| read_and_scoreA | Read any URL and return both clean Markdown AND the AIO accuracy score in one call. The recommended default for most AI workflows — gives both content and quality assessment together, so the AI agent can decide how much to trust the content. |
| batchA | Process many URLs in ONE call (billed as one request) — so you do not spend a credit per URL. Give either "urls" (an explicit list, up to 50) or "site" (a base URL whose pages are auto-discovered via sitemap). "mode" picks what to do per URL: "read" (Markdown), "read-and-score" (Markdown + AIO trust score, default), or "extract" (JSON-LD + OpenGraph + meta + score). Use this for full-site reads or bulk URL processing. |
| map_siteA | Discover a site's URLs (from sitemap.xml, falling back to on-page links) without reading them. Fast and cheap — use it to plan which pages to read or crawl next. |
| extract_dataA | Extract the structured data a page already declares — JSON-LD (schema.org), OpenGraph cards, and meta tags — plus the AIO trust score. Deterministic, no AI: returns only data present in the page. Use for fast, reliable facts (prices, products, articles) when the site publishes structured data. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 6 tools
Several tools overlap in purpose: read_url and read_and_score both retrieve content, with the latter adding a score; batch can perform both reading and extraction, duplicating read_url, read_and_score, and extract_data for bulk use. This creates ambiguous boundaries, especially when deciding between single-URL versus batch tools or whether to use read_url versus read_and_score.
Most tools follow a verb_noun pattern (read_url, score_url, map_site, extract_data), but read_and_score is a verb-verb phrase and batch is a single word without a clear verb_noun structure. The mix of conventions is still readable but not fully predictable.
With 6 tools, the server is well-scoped for its domain of URL reading, scoring, and extraction. Each tool contributes to the overall workflow without being overwhelming, and the count is within the ideal 3-15 range.
The tool surface covers the full lifecycle: map_site for discovery, read_url/read_and_score for content retrieval, score_url for quality assessment, extract_data for structured data, and batch for bulk processing. No essential operations are missing for the stated purpose of AI-ready web content extraction and analysis.