LeraLink Development Update

Over the past month, LeraLink has moved through one of its most substantial periods of development so far. The repository recorded 319 commits during this period, but the important change is not the number of commits. Several previously separate parts of the platform are beginning to form a coherent knowledge infrastructure: authoring, multilingual knowledge, identity, attribution, editorial governance, discovery, events, sharing, and now publication-grade output.

A major part of the work concerned the structure of knowledge itself. LeraLink gained a universal knowledge-organisation environment and continued consolidating the relationships between Pieces, Blocks, projects, collections, sources, people and other governed entities. Collections and structured works now behave more consistently as knowledge objects rather than as arbitrary containers. Metadata, covers, contributors, languages and catalogue records increasingly travel with the work they describe. Continuous reading, collection navigation and related-content mechanisms were also rebuilt around this structure.

Knowledge discovery became a real platform capability. Search now works across Pieces, Blocks, Projects, People and Sources, while respecting tenant boundaries, publication state and audience access. Multilingual search received particular attention: Persian normalization, ZWNJ variations, Arabic/Persian character differences and translated versions are handled as part of the indexing model rather than as superficial interface concerns. A translated Piece is not treated as an unrelated duplicate; it remains another linguistic expression of the same knowledge object. Alongside search, the knowledge index and related-reading system gained recovery, observability and backfilling mechanisms, preparing the ground for richer semantic discovery.

The authoring and translation environment also became considerably more robust. Document import and Piece portability were repaired and extended, table Blocks became first-class content, external links and embedded material received proper reader behaviour, and audio became a native Block rather than an attachment around the edges of the editor. Text and list behaviour was consolidated, highlighting was redesigned so that it survives into other representations, and citations and source relationships were made more resilient across translation. Translation governance now distinguishes between languages that may be translated into, languages publicly advertised by a tenant, stale translations and corrected translations. This matters because multilingual knowledge cannot be managed reliably if language is treated merely as a UI preference.

Another large body of work concerned people, authorship and attribution. LeraLink now has a governed person-identity foundation capable of projecting public profiles from credited work. A person can be represented consistently across the places where their work appears, while tenant-specific presentation and access remain separate concerns. Contributor roles were unified around a governed vocabulary instead of several incompatible role lists scattered through the system. Multi-author work can now identify multiple contributors correctly across reading surfaces, feeds and sharing. Just as importantly, the architecture increasingly separates credit from permission: being an illustrator, interviewee, translator, supervisor or contributor to a work does not automatically mean receiving editorial access to it.

The editorial and access model was strengthened at the same time. Review locks, revision-feedback cycles, controlled publication revisions, block provenance, granular Personal Studio permissions and clearer submission surfaces were introduced or refined. Tenant people-management and authorization were consolidated around a canonical access model, while invitation, membership and onboarding flows received repeated security and usability corrections. Operational work during the month also covered production readiness, backup recovery, authorization verification and deployment safeguards. Much of this work is invisible in the interface, but it is essential if LeraLink is to support independent organisations rather than a single centrally managed publication.

A new Events domain was added in the final part of the month. Events can now be authored as governed entities with participation, registration and live-experience capabilities, while upcoming events can appear through LeraLink’s discovery surfaces. Sharing was designed specifically for events rather than reusing generic Piece cards: event cards can communicate the date, location or online status and other essential facts before somebody even opens the link. Participants and organisers also have a dedicated visual treatment, with multilingual biographical information and individually selectable profiles.

The public-facing platform received extensive work in parallel. Share Studio, responsive social cards, source presentation, mobile navigation, tenant typography and branding controls, public SEO, first-paint performance and edge caching were repeatedly refined. The goal is increasingly that a Piece should remain recognisably the same knowledge object as it moves from the editor to the reader, search results, collections, social media or another language, while each surface can represent it appropriately.

The most recent development is the foundation of the LeraLink Publication Engine. This introduces a separate production layer for turning governed knowledge into durable publication artifacts without making the publication engine the owner of the underlying knowledge. Production settings can inherit from engine, theme, tenant, project and individual Piece levels, while capabilities can be selectively entitled per tenant.

The engine already produces real PDFs through an isolated rendering pipeline. It supports versioned publication themes, language profiles, covers, running headers, bookmarks, endnotes, bibliographies, controlled assets and fonts, private artifact storage and access policies for viewing and downloading. Persian and RTL publication are being treated as native production requirements: Persian digits, ZWNJ, Arabic/Persian character normalization, Latin runs inside Persian text, Jalali dates, right-bound books and RTL pagination behaviour are handled inside the language profile rather than patched into individual documents.

The Publication Engine is intentionally modular. Its capabilities remain dormant where they have not been entitled, making it possible for one LeraLink installation to receive only basic downloadable PDF production while another may eventually use a much richer publishing environment. This follows a broader architectural principle that has become clearer throughout the month: tenants should share the same knowledge infrastructure without being forced into the same institutional model, workflow or feature set.

Seen together, the work of the past month marks an important shift in LeraLink’s development. The platform is becoming less a collection of publishing features and more an infrastructure in which knowledge can be structured, authored, translated, credited, governed, discovered, discussed, shared and rendered into new forms without losing its identity or provenance.

There is still substantial work ahead, particularly around the Publication Engine, semantic discovery, richer media production and the interfaces through which these capabilities become simple enough for everyday use. But the underlying pieces are increasingly connected. That connection, more than any individual feature, is the most important development of this month.