Structured authorial identity that travels with the work across feeds, search, the fediverse, and AI from one source of truth.

This repository ("Author Identity") houses research and planning documents for an experiment in exposing richer, more textured WordPress user metadata that supports a more meaningful conception of authorship on the web.

It also includes the Byline Feed WordPress plugin, which implements the Byline extension vocabulary plus additional author-identity output channels. A normalized author-data layer in WordPress drives multiple outputs.

The mental model: WordPress as a Personal Data Server for authors. Ultimately, with a federated identity, any WordPress user's profile can become their "everything" folder for any content across the web that they're associated with through its production and/or ownership. Most commonly, this association is one of authorship. The Byline Feed plugin makes WordPress speak the open web's existing and emerging formats for asserting authorship. Output channels are reactive to the normalized author data — none of them own it.

This repository has two related parts:

1. A documentation set defining the vision, protocol landscape, implementation strategy, and quality gates for portable author identity in WordPress.
2. A WordPress plugin, byline-feed, that implements the MVP portion of that strategy.

• Normalize authorial user data from core WordPress, Co-Authors Plus, PublishPress Authors, and the Human Made Authorship plugin.
• Emit structured Byline metadata in RSS2, Atom, and JSON Feed.
• Expose new authorial user metadata like byline:perspective for feed consumers.
• Accommodate new authorial user metadata like byline:perspective in WordPress data model and user interface.
• Emit fediverse:creator meta tags in HTML heads for singular content by authors with configured fediverse handles.
• Emit multi-author JSON-LD Article + Person output for singular content, including Yoast SEO and Rank Math schema integration modes.
• Enrich Yoast SEO and Rank Math schema output when present, or emit standalone multi-author JSON-LD when no schema-owning SEO plugin is active.
• Preserve all standard feed elements so Byline output remains additive.
• AI-consent signaling: per-author and per-post consent, feed-level and denied-item rights metadata, robots meta for denied posts, TDMRep headers, ai.txt, and admin-side audit logging.

• Molongui adapter support
• Pre-1.0 Byline spec alignment and terminology cleanup

Longer-range identity work such as did:web: remains in the vision/research layer, not the active roadmap. Current work focuses on the output and adapter tranches without expanding the active roadmap into broader identity-framework work.

For a maintained repository tree, see docs/codebase-map.md.

From byline-feed/:

composer install
composer lint
bash bin/run-phpunit-local.sh

Alternative when you already have a WordPress test suite and database configured:

WP_TESTS_DIR=/tmp/byline-wp-tests composer test

Use bash bin/run-phpunit-local.sh for the default local path on this repo because it provisions a disposable Docker MySQL instance and avoids stale machine-specific socket config. Use composer test directly only when you already have a valid WP_TESTS_DIR plus database configuration in place.

From byline-feed/:

npm install
npm run build

Development-tooling note:

• current npm advisory state is clean after targeted dependency overrides
• npm run start still uses development-only tooling and should be treated separately from shipped plugin runtime behavior
• current policy and maintenance expectations are documented in SECURITY.md

The current Playground strategy is intentionally two-tiered:

• playground/output-demo/ is the stable source-of-truth bundle for demonstrating shipped outputs
• a later adapter-demo blueprint will showcase real Co-Authors Plus and PublishPress Authors integration without burdening the primary public demo

The output-demo README includes direct inspection routes for:

• feeds
• singular fediverse:creator and JSON-LD output
• denied rights-signaling routes
• /ai.txt

The public Playground demo runs the actual byline-feed plugin code, but uses synthetic fixture authors/content via a small demo mu-plugin so the inspection routes stay deterministic.

Use the output-demo bundle locally with:

npx @wp-playground/cli@latest server --blueprint=playground/output-demo/blueprint.json

Build a shareable snapshot ZIP with:

playground/bin/build-output-demo-snapshot.sh

Refresh the published public demo assets with:

playground/bin/publish-output-demo.sh

The public Try in Playground CTA now targets the stable codex/playground-assets blueprint URL. Each refresh republishes that stable blueprint while pinning the installed plugin to a fresh immutable source tag for the current commit. Local snapshot ZIPs remain useful for offline sharing, archived demos, and reproducible local inspection.

• Pre-release: Byline Feed 0.1.0-rc3
• Feedback and stabilization: issue #17

Release discipline:

• build the distributable plugin zip with byline-feed/bin/build-plugin-zip.sh
• update CHANGELOG.md when shipping real releases
• draft public release notes using RELEASE_NOTES.md
• include the AI-assistance disclosure note in release notes when AI materially affected the release

Key architectural rules carried through the docs and plugin:

• Byline output is additive and must not replace standard feed elements.
• All output channels consume the same normalized author contract.
• Optional upstream plugin integrations remain optional adapters, not hard dependencies.
• Scope is intentionally constrained to the defined work packages.

This repository includes material changes produced with AI assistance (OpenAI Codex and Anthropic Claude).

Preferred attribution model:

• Repository ownership and accountability remain with the human maintainer.
• Commits with substantial AI-generated changes may include an Assisted-by: trailer naming the tool used.
• AI assistance is disclosed explicitly rather than represented as human co-authorship.

Practical note:

• GitHub's contributor graph reflects commit author identity, not assistance trailers.
• AI tools are disclosed in repository documentation and commit trailers, but are not represented as separate GitHub contributor accounts.

See also:

• CONTRIBUTING.md
• CHANGELOG.md
• RELEASE_NOTES.md