Free Web Image Search

Find educational and reference images from the open web and insert them into Obsidian as remote URLs.

The plugin inserts remote image URLs and does not download selected images into your Obsidian vault.

Installation

Download the release ZIP (or the three release assets) and install them to:

YOUR_VAULT/.obsidian/plugins/free-web-image-search/

Enable Free Web Image Search under Settings → Community plugins. The base plugin requires no user API key, credit card, billing account, mandatory login, AI service, telemetry, or subscription.

V3 capabilities

V3 retains the V2 deterministic search pipeline, provider failure isolation, pagination, relevance ranking, deduplication, context search, preview, favorites, history, and remote-image health checks. It adds a larger normalized source ecosystem, source-aware provider selection, rights-aware filtering, a reusable IIIF URL layer, and specialist chemistry and structural-biology providers.

Only text explicitly used as a search query is sent to the selected provider. Surrounding note content, vault files, analytics, and telemetry are not uploaded. Context-search commands intentionally send the selected text, heading, note title, or paragraph as the query.

First-class public providers

The following sources are not silently scraped or falsely represented as working: Gallica/BnF, Wellcome, Rijksmuseum, Getty, National Gallery of Art, Walters, USGS, and NOAA. Their current no-key image endpoints and direct-image semantics require additional endpoint-specific verification before being enabled. Credentialed sources such as Smithsonian, Europeana, Harvard, DPLA, NYPL, Flickr, Pexels, Unsplash, and BHL are not required by the base plugin.

IIIF support

The reusable buildIIIFUrl() helper supports provider-supplied IIIF services with full images, width-based resizing, regions, rotation, quality, and format selection. The Art Institute provider uses its official IIIF service and exposes info.json. The URL is built from provider image identifiers, not from arbitrary scraped HTML.

Search and source-aware modes

Search modes include General, Educational, Diagram, Scientific, Mathematics, Physics, Chemistry, Biology, Geography, History, Maps, Architecture, Engineering, Anatomy, Astronomy, and Military History. Queries are expanded deterministically; no AI API is involved.

The source picker includes individual V3 sources. Specialist searches work naturally with glucose, benzene, aspirin, hemoglobin, Saturn, maps, diagrams, and reference imagery. Context commands search selected text, current heading, note title, or current paragraph without uploading surrounding content.

Rights and access-cost distinction

Free / no-user-key means the API can be used without user credentials. It does not mean every returned image is copyright-free. Cards and previews distinguish returned image-rights information such as Public Domain, CC0, Creative Commons, Open Access, Rights information available, or Rights information unknown.

The Open license only filter keeps results only when the provider returned reliable rights/license/public-domain metadata. It does not infer rights from an institution name. Licenses and license URLs are never invented.

Result actions and insertion

Result cards provide preview, insert, copy direct URL, copy Markdown, favorite, open source, context menu, and keyboard actions. Preview displays title, creator, date, collection, dimensions, format, license, rights status, source, direct URL, and IIIF information when returned.

Default insertion remains:

![Alt text](https://example.org/direct-image-url.jpg)

Optional settings support alt-text prompts, captions, compact/full attribution, and open-license filtering. URLs are limited to http:// and https://. The plugin never uses attachment upload, never writes image binaries to the vault, and never substitutes a local copy.

Remote image health

Manage Remote Images and Check Remote Images scan Markdown remote-image links in the current note, current folder, or entire vault. Results include note, line, alt text, URL, HTTP status, and detected MIME when available. Checks use bounded concurrency and never replace links automatically.

Provider health and limitations

Provider errors are isolated and shown alongside successful results. Public services can rate-limit, change endpoints, reject requests from a network edge, or block HEAD requests. Metadata differs by provider. A provider may return a source page and thumbnail but no stable original; only a validated returned URL is used. Some image hosts may block hotlinking or later remove assets.

In this build environment, the Library of Congress API may return HTTP 403 through the sandbox network edge even though the provider implementation uses its official public API. It is therefore isolated and reported as unavailable rather than hidden or misrepresented.

Privacy and security

No surrounding note content is uploaded. Only text explicitly used as a search query is sent. There is no AI integration, analytics, telemetry, account requirement, or credential collection. Unsafe URL schemes such as javascript:, data:, file:, and chrome: are rejected. Provider metadata is rendered with safe text APIs.

Development and verification

npm install
npm run build

V3 uses TypeScript as the source of truth and esbuild for the generated main.js. The final ZIP contains the compiled plugin, source, stylesheet, manifest, package metadata, TypeScript configuration, README, and data-shape reference file. No image files are included in the source repository or plugin package except the generated JavaScript release asset; selected web images are never copied into the vault.

Changelog

3.0.0

• Added Wikipedia, Art Institute of Chicago, Cleveland Museum of Art, SMK, PubChem, and RCSB PDB providers.
• Added normalized rights states, public-domain flags, IIIF metadata, and IIIF URL generation.
• Added source-aware no-key provider selection and open-license filtering.
• Added source-specific metadata details in cards and preview.
• Added specialist chemistry and structural-biology image search.
• Preserved remote-only insertion, deterministic ranking, pagination, health checks, favorites, history, and provider failure isolation.

Release checklist

For release 3.0.1:

1. Run npm install and npm run build.
2. Upload main.js, manifest.json, and styles.css from release-assets/3.0.1/ to the GitHub release tagged 3.0.1.
3. Keep main.js out of the source repository; it is ignored by .gitignore.
4. The repository root contains README.md, LICENSE, and manifest.json for Community Directory review.

The plugin uses public web APIs and image hosts. Network use is disclosed because provider APIs and the selected remote image URL must be reachable for search, preview, and inserted images to work. Rights information is provider-supplied metadata; a search result is not automatically copyright-free.