On this page
Archives need two things: structure and durability. Wire generates static HTML from markdown files. No database to corrupt. No CMS to become unsupported. No vendor lock-in. Your archive is a directory of text files and HTML pages that will be readable in 50 years. The National Archives 2022-2026 Digital Preservation Strategy emphasizes format sustainability and data integrity as core principles for federal records. Static HTML is the most sustainable web format that exists.
The Pain Point
Museums, libraries, and organizations build archives on Omeka, WordPress, or custom CMS platforms. A decade later, the platform is unsupported, the PHP version is end-of-life, and the database export is a proprietary format. The content survives only if someone migrates it. The Smithsonian maintains born-digital archives with fixity markers and redundant copies. Wire's output is plain HTML that can be archived, copied, and verified trivially.
The metadata problem is equally severe. The Library of Congress tested AI models against 23,000+ ebooks and found no model met a 95% F1 threshold for complex cataloging tasks. Only LCCN identification cleared the bar. Automated cataloging at scale remains unreliable without human-enforced structure. Wire takes the opposite approach: instead of trusting AI to generate metadata after the fact, it refuses to build without validated metadata upfront.
How Wire Fits
Wire produces the output that archivists need. The site/ directory is the complete archive: static HTML with no server-side processing, no database, no runtime dependencies. Copy it to any storage medium and it works.
Every record carries structured metadata in YAML frontmatter: title, description, created date, author, tags, and sources. Wire validates all of it on every build. Missing a title? BUILD REFUSED. No created date? BUILD REFUSED. The same strictness that frustrates content marketers is exactly what archivists require.
Citation trails are enforced by Wire's styleguide layer. Every factual claim links to its origin. Provenance is not an afterthought; it is a build requirement. This matters because archives without citation history become unverifiable over time.
Wire also generates machine-readable outputs automatically. search_index.json provides full-text search as a static JSON file, usable without a server. llms.txt creates a machine-readable inventory of every page, so AI assistants can navigate and query the archive. The sitemap and RSS feed track new additions. All of these are static files archived alongside the content.
In December 2025, OCLC shipped AI-assisted cataloging to WorldShare Record Manager, generating metadata suggestions for WorldCat records. This signals that institutional cataloging is moving toward AI augmentation. Wire fits this shift: it enforces the structure that AI-generated metadata needs to be useful, catching errors and gaps that models miss.
What to Customize
wire.yml
Organize by collection. Each collection becomes a section in your navigation, with its own index page and individual record pages:
nav:
- index.md
- Collection A:
- Overview: collection-a/index.md
- Collection B:
- Overview: collection-b/index.md
Frontmatter for Archival Records
Extend standard frontmatter with tags for archival metadata. Wire validates that every record has a title, description, and created date. Tags enable faceted browsing across collections:
Components
Use :::cards for collection browsing, where each card links to a record or sub-collection. Use :::stats for collection metrics like total items, date ranges, and provenance regions. Use :::tabs to offer multiple views of the same collection: by date, by type, or by provenance.
Compared to Other Wire Use Cases
A digital archive shares DNA with a research archive, but the goals differ. Research archives prioritize citation relationships and bibliography management. Digital archives prioritize long-term preservation and catalog completeness. Wire handles both, but the frontmatter conventions and navigation structure reflect different access patterns.
A product directory uses similar structured metadata and comparison features, but directories expect frequent updates and competitive monitoring. Archives prioritize stability. Once a record is cataloged correctly, it should not change unless new provenance information emerges.
For organizations that also need to surface archive content to AI agents, replacing RAG with static content explains how Wire's llms.txt and search_index.json outputs feed AI systems directly, without building a retrieval pipeline.
What Wire Gives You for Free
Sitemap, RSS for new additions, JSON-LD structured data, full-text search, llms.txt, responsive layout for desktop and mobile browsing, and image dimension validation. All outputs are static files archived alongside the content. No external service dependencies.
Limitations
Wire does not handle IIIF image viewers, 3D model rendering, or audio/video streaming. For rich media, Wire serves as the catalog and metadata layer. Link to external media servers for high-resolution viewing. Wire also does not generate Dublin Core or METS/MODS XML natively; if your institution requires these formats, you would export from the structured frontmatter using a post-build script.
Quick Start
Run python -m wire.build from your site directory. Wire reads your wire.yml, validates every record's frontmatter, generates the static archive, and runs 91 build checks. The output in site/ is a self-contained directory of HTML, CSS, JSON, and XML. No dependencies, no runtime, no expiration date.
## Minimal wire.yml for an archive
site_name: My Archive
site_url: https://archive.example.org
nav:
- index.md
- Manuscripts:
- Overview: manuscripts/index.md
- Photographs:
- Overview: photographs/index.md