Agent View
The site as an AI agent gets it, beside the site as a person sees it. What it can read, what it has to cut away, and what it never sees.
Skip to main content
Cory Dransfeldt[email protected]~/coryd.dev ~ / c o r y d . d e vPostsMusicWatchingReadingLinks
Public applications
- coryd.dev: this site.
- Mastodon: my fediverse instance.
- Bluesky PDS: my AT Protocol data server.
Personal services
- Forgejo: git hosting and CI runners.
- Music: Navidrome, with AudioMuse-AI for playlists and similar artists.
The rest of the page is in full screen.
## Public applications - [coryd.dev](https://www.coryd.dev/): this site. - [Mastodon](https://follow.coryd.dev/): my fediverse instance. - [Bluesky](https://bsky.app/profile/coryd.dev) PDS: my AT Protocol data server. ## Personal services - [Forgejo](https://forgejo.org/): git hosting and CI runners. - Music: [Navidrome](https://www.navidrome.org/), with [AudioMuse-AI](https://github.com/NeptuneHub/AudioMuse-AI) for playlists and similar artists. - Video: [Jellyfin](https://jellyfin.org/) - Bookmarks: [linkding](https://linkding.link/) - Notes sync: [CouchDB](https://couchdb.apache.org/) for [Obsidian LiveSync](https://github.com/vrtmrz/obsidian-livesync). - Contacts: [Radicale](https://radicale.org/) - Stack: my reading app for feeds, bookmarks, books, audiobooks and highlights. - Concert tracker: a one page app that compares the artists in my Navidrome library against Ticketmaster's API to show me concerts in my area. ## Infra - [Headscale](https://headscale.net/): the control server for my Tailscale network. - [CoreDNS](https://coredns.io/): internal DNS, run on two hosts. - [Gatus](https://gatus.io/): uptime monitoring. - [Caddy](https://caddyserver.com/): reverse proxy and TLS on every host. - [Postgres](https://www.postgresql.org/) and [pgcat](https://github.com/postgresml/pgcat): databases and connection pooling. - Deploy queue: [webhook](https://github.com/adnanh/webhook)-driven deploys. - Backups: encrypted to [Backblaze B2](https://www.backblaze.com/cloud-storage).
No h1: Give each one h1 that says what the page is.
The figures
100 pages, set against the 168,007 sites with figures and 197 in news.
- Readablehigher is better
- 82%82 of 100 pages
- All 168,007 sitesBetter than 25%197 news sitesBetter than 26%
Pages whose own text is in the HTML the server sends, enough of it for their kind.
- Cleanhigher is better
- 74%middle page
- All 168,007 sitesBetter than 55%197 news sitesBetter than 57%
Of the words before script, the share that are the page and not menus, footer and forms. 19 pages are not measured, with no <main> or no words.
- Costshown, never scored
- 22,115tokens
- About 221 a page
To read every page once. A long page that answers well is not worse for it. Tokens are estimated at about 4 characters each.
How its pages spread on Clean
81 pages measured. Not measured: 1 with no <main> or no words and 18 shells with no words before script.
Against its rivals
The sites nearest it by what they do, with figures. Close means similar work, not better.
| Readinghigher is better | Costnot scored | ||
|---|---|---|---|
| Site | Readable | Clean | Tokens a page |
| coryd.devThis site, 100 pages | 82% | 74% | 221 |
| ritiksahni.com69 pages | 93% | 87% | 524 |
| damien.zone100 pages | 53% | 96% | 926 |
Second of 3 on Readable, third on Clean.
AI crawler access
What robots.txt says to each, as read on October 23, 2026. 1 of 8 may read the site.
| Crawler | robots.txt says | Used for | The line that decides |
|---|---|---|---|
| GPTBotOpenAI | Blocked | Training models | User-agent: GPTBot / Disallow: / |
| OAI-SearchBotOpenAI | Blocked | ChatGPT search | User-agent: OAI-SearchBot / Disallow: / |
| ChatGPT-UserOpenAI | Blocked | A page a person asks about | User-agent: ChatGPT-User / Disallow: / |
| ClaudeBotAnthropic | Blocked | Training models | User-agent: ClaudeBot / Disallow: / |
| Claude-SearchBotAnthropic | Blocked | Claude’s search | User-agent: Claude-SearchBot / Disallow: / |
| PerplexityBotPerplexity | Blocked | Perplexity search | User-agent: PerplexityBot / Disallow: / |
| Google-ExtendedGoogle | Blocked | Training Gemini | User-agent: Google-Extended / Disallow: / |
| bingbotMicrosoft | Not named | Bing and Copilot | Falls to User-agent: * |
7 of 8 are kept out. A crawler kept out never reads a page, however well it is made.
Every page
Worst first. Each opens in Agent View at the top.
20 pages of 100 lose something. Pick a loss to see what to do about it.Open the action plan
Show where it stands
A card, a badge and a line, each with the figures.
A badge for the site
<a href="https://mttr.si/agent-view/coryd.dev"><img src="https://mttr.si/agent-view/coryd.dev/badge.svg" alt="Agent View: 82% readable, 74% clean" height="20"></a>
The figures in one line
coryd.dev in Agent View: 82% of pages readable, better than 26% of news sites; 74% clean; 22,115 tokens to read all 100 pages. mttr.si/agent-view/coryd.dev
What to do next
Fix the pages, let an agent do it, and read the site again to see the figures move.
Fix what agents miss
- Put the text of 18 pages in the HTMLReadable
- Give 2 pages a main headingClearer
- Wrap 1 page in <main>Clean measured
Connect your agent
Claude Code, Codex or Cursor reads the site’s fixes over MCP, fixes a page in your code, has mttr read it again, and checks the fix landed.
https://mttr.si/mcpSet up an agent
Claim the site
Prove it is yours with a DNS record or a file. Your agent can then reach its reads and jobs, and set each change to wait for your approval.
Claim the siteDrawn from the HTML mttr fetched on its read of October 23, 2026, not a screenshot, and only from the 100 pages read, where the read stopped at its page limit. An assistant’s own reader may differ. Tokens are estimated at about 4 characters each. A site an agent reads well is not a site an assistant cites.