Guide · Obsidian

How do I get my Obsidian plugin found in AI search?

Updated 2026-11-11 · kagentx

Short answer

Your plugin has only three places on the public web: the plugin page in the community directory (the title comes from name, the meta description from description, and the body is basically the README), the GitHub repo home page, and community-plugins.json with just five fields. Tested 2026-11-11: obsidian.md's robots.txt allows everything, and GitHub blocks raw files, directory trees and similar sub-paths for AI crawlers but not the repo home page. So what you need to do is write the description and the README first screen clearly.

Which public pages carry an Obsidian plugin's information?

Three places: your plugin page in the community directory, your GitHub repo home page (which is your README), and the community-plugins.json list. On 2026-11-11 we took one sample plugin (Git): obsidian.md/plugins?id=obsidian-git 301-redirects to community.obsidian.md/plugins/obsidian-git, and that page's HTML directly contains the title, description, canonical, h1 and README text, with no script execution needed.

community-plugins.json had 10,232 entries that day, each with only five fields: id, name, author, description and repo. No README, no tags, no screenshots. So all AI can read from this list is your name, your author name and a one-line description.

Where does each of the five fields come from and where does it appear?

FieldSourceWhere it appears (tested 2026-11-11)
idid in manifest.jsonThe plugin page URL community.obsidian.md/plugins/<id>; must be unique and cannot contain obsidian
namename in manifest.jsonThe plugin page <title> (“Git - Obsidian Plugin”) and <h1>; also used by in-app search
authorauthor in manifest.jsonThe author name on the plugin page; used by in-app search
descriptiondescription in manifest.jsonThe meta description of the plugin page, identical to the sentence in community-plugins.json
repoThe GitHub repo you submittedWhere the plugin page reads the README from; also the other page AI can see (the repo home page)

The official docs (Manifest page) list name, author, description, id, isDesktopOnly, minAppVersion and version as required. The name should be short and descriptive, must not contain “Obsidian”, and a plugin name must not contain “Plugin”. The docs give no length limit for description (not verified). From obsidian-developer-docs, original text read on GitHub on 2026-11-11.

What does the README do on the plugin page?

The official submission docs say README.md should describe the plugin's purpose and usage, an excerpt is shown on the public listing page, and relative links and images are rewritten to point at your repo. In the sample page we fetched, the visible text is about 10,000 characters and contains README sections (feature list, documentation link, mobile notes).

In other words, the body of the plugin page is basically your README. If the first paragraph of your README is vague, what AI reads on the plugin page and on the repo home page is the same vague text. How to write it: Will AI search cite my GitHub README?; the overall approach for open-source projects: How do open source projects get recommended by AI search? Six places beyond the README.

# Git (title: what it does, in the plain words people search)

One sentence: what the plugin does and for whom.
Install: Settings > Community plugins > Browse > search the name.
Features: a bullet list, one real feature per line.
Settings and commands: the exact command names.
Mobile / desktop only: say it.
Changelog: link to GitHub releases.

How do obsidian.md and GitHub set robots.txt for AI crawlers?

Siterobots.txt today (2026-11-11)What it means for AI crawlers
obsidian.md200, 23 bytes: User-agent * Allow: /No AI crawler is named; all are allowed
community.obsidian.mdNo real robots.txt: /robots.txt returns 200 but the content is an HTML page; the home page has noindex, the sample plugin page has no robots metaNo robots rules to read; home page is noindex, the plugin page itself has no noindex
github.comGPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, Claude-User, PerplexityBot and Perplexity-User share one group, Crawl-delay: 1, about 280 Disallow linesBlocks /*/raw/, /*/tree/, /*/blame/, /*/archive/, commit lists, compare pages and similar; the repo home page is not in a Disallow (read from the text, not verified with a real crawler)
github.com (Bytespider)Disallow: /Whole site blocked

So what AI crawlers can read is the plugin page in the community directory, plus the GitHub repo home page and the rendered README. What they cannot read is raw files and directory trees, which means key information placed only in a sub-file under docs/ is worse than information on the README front page. How to write robots.txt: How do you write robots.txt for AI crawlers?.

What should I do after submitting?

  1. 1. Write the first description sentence in words people search

    The description appears in both the meta description and in-app search. Write what the plugin does and who it is for, not just a feature name. For example, “Integrate Git version control with automatic backup” is easier to quote than “Git plugin”.

  2. 2. Answer “what is it, how to install, what can it do” on the README first screen

    Follow the template above. Write the install step as Settings > Community plugins > Browse, which is the question users ask.

  3. 3. Write one more sentence in the GitHub repo About

    The About text of the repo appears in the title and description of the repo home page. In the sample it differs from the directory description, so writing both gives AI two chances to read you.

  4. 4. Check the public page with curl

    Replace the id with your plugin and check that the title and description are the sentences you wrote, and that a sentence from your README can be grepped. The body text must be in the HTML; the reason is in Do AI crawlers render JavaScript?.

    # What does the public plugin page say in raw HTML?
    curl -sL -A "Mozilla/5.0" "https://obsidian.md/plugins?id=YOUR-PLUGIN-ID" | grep -io '<title>[^<]*</title>'
    curl -sL -A "Mozilla/5.0" "https://obsidian.md/plugins?id=YOUR-PLUGIN-ID" | grep -io '<meta[^>]*name="description"[^>]*>'
    curl -sL -A "Mozilla/5.0" "https://obsidian.md/plugins?id=YOUR-PLUGIN-ID" | grep -c "a sentence from your README"
  5. 5. Ask three times in ChatGPT and Perplexity

    Ask “how do I do X in Obsidian”, turn on web search, and note which plugins are recommended and which URLs are cited. Whether ChatGPT or Perplexity cites community.obsidian.md plugin pages, we have no evidence and it is not verified; you have to test it yourself.

How is Obsidian different from other plugin marketplaces?

Obsidian community directoryGitHub repo
Who generates the pageThe directory generates it from the manifest and READMEYou
Text you can changename, description (edit the manifest), READMEAbout, README, Topics
What AI crawlers can readTitle, description, README body (sample page)Repo home page and README; raw and directory trees are blocked
What the official docs do not sayThe description length limit, whether AI cites these pages—

If you build a Figma plugin, the Figma Community plugin page's robots.txt, sitemap and the vendor's listing requirements are in How do AI search tools find Figma plugins? The Community plugin page's robots.txt, sitemap and Figma's requirements.

Plugin ecosystems also include Minecraft mods and server plugins; the Modrinth and CurseForge pages we checked are covered in How do Minecraft mods get found by AI search?.

FAQ

Can ChatGPT and Perplexity read Obsidian plugin pages?

In the 2026-11-11 test, the plugin page HTML directly contains the title, description, canonical and README text, with no noindex. Whether real AI crawlers read it and whether they cite it, we have no evidence and it is not verified.

Which fields are in community-plugins.json?

Five: id, name, author, description and repo; 10,232 entries on 2026-11-11. No README and no screenshots.

Is there a length limit on a plugin's description?

The official Manifest page gives no number; not verified. Write one sentence that says what it does and who it is for.

Does GitHub block AI crawlers from reading my README?

The Disallow lines in github.com/robots.txt for the eight AI crawlers are mainly sub-paths such as raw, directory trees, commits and compare; the repo home page is not among them. This is read from the text, not verified with a real crawler.

Can I put keywords in the plugin name?

The official rule is that the name be short and descriptive, must not contain “Obsidian”, and a plugin name must not contain “Plugin”. Put keywords in the description and the README.

Do I need a separate website for my plugin?

Not required. The plugin page and README are already server-rendered body text. If you build one, make it a description page that matches the README and links back to the plugin page.

Why is the community.obsidian.md home page noindex?

We only saw noindex on the home page, not on the sample plugin page. The official docs do not explain why; not verified.

Sources

kagentx

Do AI answers
mention you?

An evaluation asks ChatGPT, Perplexity and Claude your buyers' questions (with web search on) and records who they recommend, which pages they cite and whether you are among them. Basic $1, results usually in 2–3 minutes.