Guide · Blender extensions

How do AI search tools find Blender add-ons? The Extensions platform tested, plus the official listing requirements

Updated 2026-11-14 · kagentx

Short answer

Tested 2026-11-14: the Blender Extensions platform (extensions.blender.org) has no sitemap (/sitemap.xml is 404); its robots.txt names and disallows training crawlers such as GPTBot, ClaudeBot and CCBot, but does not name OAI-SearchBot or PerplexityBot; the detail page is 59,183 bytes with the body server-rendered, a meta description that is the manifest tagline of at most 64 characters, and no JSON-LD. So the tagline, tags and detail page introduction are all you control; the vendor requires GPL-3.0-or-later for add-ons and blender_version_min of at least 4.2.0, and publishes after review. Whether AI cites these pages, we have no evidence: unverified.

Is the Blender Extensions platform open to AI crawlers?

Partly: training crawlers are named and disallowed, while search and user-triggered crawlers are not named. On 2026-11-14 we tested extensions.blender.org with curl (User-Agent Mozilla/5.0):

CheckTest result
robots.txt200, 2,749 bytes. Cloudflare-managed block: the * group has Content-Signal: search=yes,ai-train=no,use=reference and Allow: /; GPTBot, ClaudeBot, anthropic-ai, Claude-Web, CCBot, Google-Extended, Applebot-Extended, Amazonbot, Bytespider and others each have Disallow: /; at the end of the file there is another * group with Disallow: /approval-queue/ and another GPTBot group with Disallow: /
Crawlers not namedOAI-SearchBot, ChatGPT-User, PerplexityBot, Claude-SearchBot and Claude-User do not appear, so by the wording of robots.txt they fall under the * group (allowed). This is inferred from the text and not verified with real crawlers
sitemap/sitemap.xml and /sitemap-index.xml are both 404; robots.txt has no Sitemap line
JSON repository index/api/v1/extensions/: 200, 1,094,475 bytes, no login needed; 1,654 extensions: 1,357 add-ons, 289 themes, 8 asset libraries

So this site has no sitemap you can submit to crawlers; crawlers find extension pages through links on the listing pages. For the names and purposes of the AI crawlers see AI crawler user agents: a side-by-side table; for how to write robots.txt see How do you write robots.txt for AI crawlers?.

What is in the HTML of an extension detail page?

The body text, but no structured data. We fetched the EdgeFlow page extensions.blender.org/add-ons/edgeflow/ (2026-11-14, curl, no scripts run):

CheckTest result
HTTP status and size200, 59,183 bytes
title“EdgeFlow — Blender Extensions”
meta descriptionHelps adjusting mesh geometry to curved surfaces, the same as the manifest tagline
canonical and og tagscanonical present; og:title, og:description, og:image and og:image:alt present
JSON-LDNone
Visible text (scripts removed)3,524 characters, including the introduction text (“This addon adds a set of operators …”): the body is in the HTML and not rendered by script

This is the opposite of JetBrains Marketplace in How do AI search tools find JetBrains plugins? The Marketplace plugin page, tested, where the detail page is rendered by script. Here, as long as a crawler is let in, it can read the full introduction; whether it is let in depends on the rules in the previous section. For whether crawlers run scripts, see Do AI crawlers render JavaScript?.

What are Blender's official requirements for listing an extension?

The official manual (docs.blender.org, Blender 5.3 Manual, original text read 2026-11-14), “How to Create Extensions”: write blender_manifest.toml in the extension folder, build a .zip with blender --command extension build, and test it yourself with Install from Disk; uploading the .zip needs a Blender ID; after upload the extension goes into review and is published once the review team approves it. The required manifest fields:

FieldOfficial requirement
idUnique identifier of the extension
nameFull name of the extension
taglineA one-line short description, up to 64 characters, not ending with punctuation
typeOne of add-on, asset-library or theme
versionMust follow semantic versioning
maintainerMaintainer
licenseA list of SPDX identifiers; add-ons must use GPL-3.0-or-later and assets inside add-ons require CC0; for themes GPL-3.0-or-later is recommended and any GPL-compatible license is accepted
blender_version_minAt least 4.2.0
schema_versionWrite 1.0.0
schema_version = "1.0.0"
id = "edge_flow_helper"
version = "1.0.0"
name = "Edge Flow Helper"
tagline = "Fit edge loops to curved surfaces"
maintainer = "Your Name"
type = "add-on"
website = "https://example.com/edge-flow-helper/"
tags = ["Modeling", "Mesh"]
blender_version_min = "4.2.0"
license = ["SPDX:GPL-3.0-or-later"]

Among the optional fields, the ones that matter for being found are website, tags, platforms and permissions. tags must come from the official list, which has 31 for add-ons, such as Animation, Geometry Nodes, Import-Export, Modeling, Rigging and UV. permissions can be files, network, clipboard, camera and microphone, and each needs a reason of at most 64 characters without a trailing period. The manifest must not contain empty strings or empty lists; drop optional fields you do not want to fill in entirely.

A few points from the reviewers' guidelines (the Extensions moderation guidelines on developer.blender.org, original text read 2026-11-14) that relate to the listing page:

  • Check the overall quality of the images, and do not put heavy images (especially .gif) in the description, which would make the page huge.
  • The author must be the owner or maintainer of the extension.
  • The extension must be self-contained and not depend on external servers (localhost is fine).
  • No auto-updater code; pip dependencies must be bundled as wheels; anything using the network permission must check bpy.app.online_access first.

The vendor does not say how extension pages are ranked, nor how these fields affect AI citation; we have no evidence for either: unverified.

How do you make a Blender extension easier to find, step by step?

  1. 1. Write the tagline as a one-sentence definition of the function

    At most 64 characters, with no ending punctuation. It is both the detail page's meta description and the snippet in search results: say what it does, and leave out adjectives.

  2. 2. Keep the name and id consistent

    Write the name users would search for in name; the id is not shown to users, but the name part of the URL comes from the extension name, so think before renaming.

  3. 3. Choose the right tags

    Pick 1–3 tags from the official list that really apply, and do not pad.

  4. 4. Put the introduction text in the detail page

    The body on this site is server-rendered, so put the features, supported Blender versions, steps and screenshots in the detail page's introduction rather than only in external links.

  5. 5. Write a full introduction on your site or GitHub and link them both ways

    Put your official site in the manifest's website field; your site then links back to the Extensions platform page. For how to write GitHub, see Will AI search cite my GitHub README?; for the other places for open source projects, see How do open source projects get recommended by AI search? Six places beyond the README.

  6. 6. Test again after a few weeks with the same questions

    Ask ChatGPT and Perplexity “is there a Blender add-on that does such-and-such”, ask each question 3 times, and note whether you are mentioned and which URLs are cited.

What about Blender tools sold on Superhive and Gumroad?

The Extensions platform only takes open source (GPL-compatible) extensions. Tools meant to be sold usually go on marketplaces such as Superhive (formerly Blender Market; blendermarket.com now redirects to superhivemarket.com) or Gumroad. On 2026-11-14 we tested only the robots.txt of the two sites, not product detail pages:

SiteTest result
superhivemarket.comrobots.txt 200, 64 bytes, no Disallow at all in the * group; Sitemap points to an index, which holds 3 child sitemaps (lastmod 2026-11-13)
gumroad.comrobots.txt 200, 33,650 bytes, the * group only disallows /purchases/; 375 Sitemap lines; no AI crawlers named

These two sites are more relaxed toward crawlers than the Extensions platform, but we did not test whether detail pages return the body directly: unverified. For how to write on Gumroad-type platforms, see How do digital products sold on Gumroad and Lemon Squeezy get recommended by AI?. For comparable plugin stores: How do AI search tools find Figma plugins? The Community plugin page's robots.txt, sitemap and Figma's requirements, How do AI search tools find JetBrains plugins? The Marketplace plugin page, tested.

FAQ

Does the Blender Extensions platform's robots.txt block AI crawlers?

Some of them. The robots.txt tested on 2026-11-14 is 2,749 bytes: it names training crawlers such as GPTBot, ClaudeBot, anthropic-ai, Claude-Web, CCBot and Google-Extended with Disallow: /; OAI-SearchBot, ChatGPT-User, PerplexityBot and Claude-SearchBot are not named, so by the text they fall under the * group, which has Allow: / and disallows only /approval-queue/.

Does extensions.blender.org have a sitemap?

No. Tested on 2026-11-14, /sitemap.xml and /sitemap-index.xml are both 404, and robots.txt has no Sitemap line. The site does have a JSON repository index (/api/v1/extensions/, 1,654 extensions), but that is not a sitemap.

Can a crawler that does not run scripts read the body of an extension detail page?

Yes. On 2026-11-14 we fetched the EdgeFlow page (59,183 bytes); with scripts removed there are 3,524 visible characters, including the introduction text; the title, canonical, meta description and og tags are all there, and there is no JSON-LD.

Where does the detail page's meta description come from?

In our test it is the same as the manifest tagline: EdgeFlow's is “Helps adjusting mesh geometry to curved surfaces”. The vendor specifies that the tagline is at most 64 characters and cannot end with punctuation, so this one sentence is both the snippet in search results and the place to make the function clear.

What does it take to publish a Blender extension?

The official manual says: pack a .zip containing blender_manifest.toml; upload after logging in with a Blender ID; it goes into review and is published once the review team approves it. An add-on's license must be GPL-3.0-or-later, and blender_version_min must be at least 4.2.0.

Which tags should I pick?

The official manual lists 31 tags for add-ons, such as Modeling, Mesh, Animation, Geometry Nodes, Import-Export and Rigging. Pick 1–3 that really apply to the function; the vendor does not say how tags affect on-site search: unverified.

What about the robots.txt of Blender Market (now Superhive) and Gumroad?

Tested 2026-11-14: blendermarket.com redirects to superhivemarket.com, whose robots.txt is only 64 bytes with no Disallow and a Sitemap pointing to 3 child sitemaps; the gumroad.com robots.txt only disallows /purchases/ and has 375 Sitemap lines. Neither site names AI crawlers. We did not test product detail pages: unverified.

Does AI search cite pages on extensions.blender.org?

We have no verifiable source: unverified. What is certain is that the page body is server-rendered, there is no sitemap, and the site names and disallows AI training crawlers.

Sources

kagentx

Can crawlers read your pages?
Let the evaluation tell you

The evaluation checks whether crawlers can get in and whether the text is served directly, then asks ChatGPT, Perplexity and Claude (with web search on) your buyers' questions and sees whether you are mentioned. The Basic plan is $1, and results usually take 2–3 minutes.