Daily Digest

Daily Digest – 22 June 2026

2026 Capabilities Progress

  1. Page generation — Progress on AI agent infrastructure continued with improvements to layout fixture support for testing page builder agents [66][68], though critical bugs remain where AI generates components using unsupported libraries [67]. Work on component generation prompts is addressing issues with deprecated imports and hardcoded error messages [57][84].
  2. Context management — The Context Control Center merged three user experience improvements including configuration links, clearer scope form layouts, and validation preventing invalid settings [94][97][99][118][120][122]. Active work continues on scope plugin persistence, multi-vocabulary taxonomy support, and integration with document loading capabilities [115][116][98].
  3. Background agents — The ECA Interceptor submodule remains in development to enable event-driven AI workflows, but the issue has no assigned owner and saw no commits today [44].
  4. Design system integration — Canvas is developing conflict detection UI for selective publishing, with the core merge request landed but underlying conflict detection and prevention issues still open [77][79] and improved data model handling for component properties [82][89], while symmetric translation support to propagate component updates across languages remains under active review [55][80][87].
  5. Content creation and discovery — AI Search merged long-running chunk indexing for processing documents beyond 10 chunks [8][34][130] and is actively reviewing filterable RAG tools for metadata-based search refinement [123][129]. The AI module improved editor experience warnings when providers are missing [3][21] but faces quality issues with taxonomy automators corrupting HTML-encoded characters [14].
  6. Advanced governance — A new moderation guardrail plugin allowing configurable content safety checks through any moderation service is under review [2][20]. Canvas is developing conflict detection and resolution UI with detection merged but resolution still in draft, and conflict prevention issue remains open [52][58][77][79][91].
  7. Intelligent website improvements — No direct progress today.
  8. Multi-channel campaigns — No direct progress today.

Shipped

  1. AI module: Long-running content indexing support — Support for processing documents longer than 10 chunks was merged to the stable 1.x branch, though the feature issue remains open [8][34][130].
  2. Canvas: Language-aware content indexing — Backend support for language-aware indexing of ContentTemplate-rendered Canvas content was completed [60][86].
  3. Canvas: Conflict detection and resolution interface — Work on conflict detection and resolution UI is in progress, with some merge requests merged but the feature issue still open [58][77][79].
  4. Context Control Center: Configuration improvements — Three usability enhancements were delivered including a standard configure link on the modules page, clearer scope form guidance, and validation preventing conflicting administrator settings [94][97][99][118][120][122].
  5. AI Search: Enhanced filtering capabilities — An attempt at long running chunk indexing for the stable branch was merged without breaking compatibility, and a user question about metadata filtering in chatbot search was resolved [124][126][130].

Ongoing

  1. AI module: Content moderation guardrail — A new plugin allowing administrators to configure any moderation service as a content safety check is under review [2][20].
  2. AI module: Data quality fixes — Two issues affecting editors are being addressed: taxonomy automators losing valid terms when AI returns HTML-encoded characters, and missing token usage data during streaming responses [11][14].
  3. Canvas: Symmetric translation support — Work continues to ensure component updates propagate correctly across all language versions [55][87].
  4. AI translate: HTML preservation bugs — The module is not reliably preserving formatting when translating rich text content, with markup being corrupted or improperly inserted [137][138].
  5. Context Control Center: Beta release preparation — The team is recruiting beta testers, creating instructional videos, and completing scope plugin improvements ahead of the release candidate [102][105][115].

AI (Artificial Intelligence)

The AI module made significant progress on infrastructure and content quality features over the past day. The team merged support for handling large content items, enabling the search system to process documents longer than 10 chunks without breaking—critical for sites indexing full-length articles or documentation. This brings the stable 1.x branch closer to feature parity with the development version while maintaining compatibility.

Development work continued on several user-facing improvements. A new moderation guardrail plugin is in review that will let administrators use any moderation service as a content safety check through simple configuration. Contributors also identified and are fixing two quality issues affecting editors: taxonomy automators losing valid term values when AI models return HTML-encoded characters (like "&" for "&"), and missing token usage data when streaming responses in certain execution contexts.

The team improved onboarding by adding warnings when no AI provider is installed in the CKEditor integration, and resolved a namespace collision between a deprecated bundled module and its standalone project replacement. Several contributors addressed permission clarity issues for the AI Assistant chatbot feature.

How can I help on this project?

  • Prioritize testing and feedback on the moderation guardrail plugin to validate whether the configuration approach meets real-world content safety requirements.
  • Clarify the roadmap for chatbot permissions: confirm whether consolidating access controls is the strategic direction or if the current dual-permission model serves distinct use cases.

View project on GitLab: ai

Drupal AI Initiative

The AI Initiative saw significant planning activity over the past 24 hours, focused on two main areas: improving the user experience for discovering AI tools, and organizing an upcoming Enterprise AI Summit event.

A UX meeting on June 15 was documented, highlighting progress on a new searchable directory of AI modules and recipes. This directory will help site builders find the right AI tools more easily through improved filtering and badging that indicates which modules are officially supported by the initiative.

The team also created several planning tasks for the Enterprise AI Summit, including updating sponsor information on the event website (with three gold and multiple silver sponsorships secured), improving mobile accessibility, and recruiting event volunteers. Additional tasks cover sponsor lead collection systems and outreach to business organizations for event partnerships.

These activities represent early-stage event planning and product development work, with multiple tasks still seeking contributors.

How can I help on this project?

  • Support decision-making on partnership strategy with business organizations (ThomsonReuters, CIONet) for the Enterprise AI Summit to expand audience reach.
  • Allocate resources to help recruit Dutch-speaking event volunteers for registration and attendee support.
  • Consider funding for social media promotion to increase event visibility and attendance.

View project on GitLab: ai_initiative

AI Integration - ECA

During this period, work continued on the Add AI ECA Interceptor submodule, which will allow administrators to monitor and respond to AI system events without writing code. This feature enables business rules to be configured around AI interactions—such as logging when AI services fail, tracking usage patterns, or implementing approval workflows when content generation occurs. The capability is part of the project's roadmap toward a stable 1.0.0 release.

No code was merged or committed in the past 24 hours. The issue remains open and unassigned, which is normal for a module in active development working toward its first stable version.

How can I help on this project?

  • Consider allocating developer resources to advance the AI ECA Interceptor feature, which currently has no assigned owner.
  • Support prioritization discussions to clarify which capabilities are essential for the 1.0.0 release versus future enhancements.

View project on GitLab: ai_integration_eca

OpenAI Provider

The OpenAI Provider module saw development progress focused on expanding functionality for image manipulation. Marcus Johansson (Marcus_Johansson) opened a new merge request to add image-to-image capabilities, representing 248 lines of code changes. This enhancement will enable users to generate new images based on existing images, expanding the module's creative and practical applications beyond text-based interactions.

No critical issues or blockers were reported during this period. The module continues its development trajectory with no commits merged in the last 24 hours, indicating that submitted work is undergoing review before integration.

How can I help on this project?

  • Allocate reviewer capacity to evaluate and approve the pending image-to-image functionality, which will accelerate delivery of new capabilities to users.
  • Clarify strategic priorities for AI feature development to help contributors focus efforts on the highest-value enhancements.

View project on GitLab: ai_provider_openai

AI Agents

Activity over the past 24 hours was limited to issue tracking. A Schema error for AI Agents CKEditor plugin was logged, identifying a configuration issue where the module's text editor integration lacks proper schema definition. This causes Drupal to report errors, though it does not prevent functionality. The issue reflects ongoing quality improvements as the module matures its integration with text editing capabilities.

No code was merged or committed during this period, indicating a quiet day for development activity.

How can I help on this project?

  • Prioritize developer time to address the CKEditor schema error, which affects configuration validation quality.
  • Consider allocating resources for thorough testing of existing integrations to surface similar issues proactively.

View project on GitLab: ai_agents

Drupal Canvas

The Canvas team delivered eight improvements over the past 24 hours, focused on translation support, data model refinement, testing stability, and user interface enhancements.

The team merged language-aware indexing for ContentTemplate-rendered Canvas content, ensuring that Canvas content is searchable in the correct language. They also completed conflict detection and resolution UI for selective publishing, now available behind a development flag. This allows teams to see when multiple editors have made changes to the same content and provides a path to resolve conflicts before publishing. Ben Mullins (bnjmnm) and Ravi Maniyar (ravi.maniyar.123) led this work.

Several technical improvements were also merged: Christian López Espínola (penyaskito) and Wim Leers (wimleers) refactored internal data handling to simplify testing and maintenance. Feliksas Mazeikis (f.mazeikis) resolved a testing framework issue that was causing false failures. Ben Mullins addressed flakiness in language selector tests. Bálint Kléri (balintbrews) updated JavaScript dependencies in preparation for the 1.7.0 release and removed deprecated command-line tools that had been replaced by newer alternatives.

Active work continues on symmetric translation support, ensuring component updates propagate correctly across all language versions, and preventing publishing when conflicts are detected. Both remain in progress with merge requests under review.

How can I help on this project?

  • Confirm whether content entity reference props should remain behind a development flag or be prioritized for general availability, as the underlying work is unassigned.
  • Review the strategic direction for AI-generated components to clarify library constraints and acceptable error handling, as multiple related issues remain open without clear next steps.

View project on GitLab: canvas

Context Control Center (CCC)

The Context Control Center continues to refine its user experience and expand its capabilities ahead of the stable release. Over the past 24 hours, three improvements were merged that make the module easier to configure and use.

The team closed Add Configure Link in Modules Page, contributed by Rob Loach, which introduces a standard configuration link on the module listing page. This small change improves discoverability for administrators setting up the system. Copy and layout updates on context item scope form was also completed, clarifying help text and reducing clutter on the scope configuration screen to help users understand when context will be included. Additionally, Matt Glaman delivered Add context settings validation: max_global_items must not exceed max_items, preventing administrators from entering conflicting limits during configuration.

The project is actively working on scope plugin improvements, integration with external modules like Document Loader, and preparing for a release candidate. Sprint planning is underway for Sprint 12, with Kristen Pol leading roadmap updates and issue triage. An issue tracking Drupal 12 compatibility has been opened and is receiving automated updates.

How can I help on this project?

  • Support the team in recruiting beta testers through organizational communication channels and partnerships to accelerate feedback and stabilization.
  • Allocate resources to complete documentation and instructional videos that help new users understand configuration, reducing friction for adoption.

View project on GitLab: ai_context

Field Widget Actions

The Field Widget Actions module saw focused development activity this week on improving how automated content suggestion buttons are configured and deployed across forms.

Work continued on Generate UUIDs for list-style widget-action settings applied via config actions, which addresses a limitation preventing multiple AI suggestion buttons from being added to the same form through automated deployment recipes. Previously, buttons would collide and write suggestions to the wrong fields. The solution allows each button to receive a unique identifier automatically, matching the behavior users already experience when adding buttons manually through the interface. This work supports scalable, repeatable deployment of AI-assisted editing capabilities across sites.

Additionally, progress was made on Field Widget Actions show in the "Add New Action" unavailable tools, assigned to Joonas Meriläinen (mErilainen). This issue resolves confusion where action buttons appear as available options in dropdown lists but cannot actually be added due to missing dependencies, improving the user experience when configuring forms.

How can I help on this project?

  • Prioritize testing and review resources for the existing merge request on UUID generation to accelerate delivery of multi-button deployment capability.
  • Support coordination between this module and broader AI initiative teams to ensure alignment on configuration patterns and deployment workflows.

View project on GitLab: field_widget_actions

AI translate

The AI translate module saw five new issues raised during this period, signaling active user engagement but also highlighting areas where the module's current capabilities need refinement. All activity was in the form of new issue creation; no development work was merged.

Two bug reports point to a core quality concern: the module is not reliably preserving HTML formatting when translating rich text content. Users report that markup such as paragraph and anchor tags are being corrupted or improperly inserted during translation, which would degrade content quality for sites using formatted fields. Both issues have been accepted and are awaiting work.

On the feature side, a proposal for batch translation suggests the current approach is inefficient—translating one field at a time leads to redundant API calls and higher costs. The assigned contributor is exploring aggregating multiple fields into a single request to reduce overhead. A broader architectural question has also been raised about refactoring the translation workflow to support agent-driven translation and better code reuse across different access points.

A user question about the "Chat proxy to LLM" provider remains unanswered, indicating a need for clearer documentation or community response.

How can I help on this project?

  • Prioritize development resources to address the HTML preservation bugs, as these directly affect content quality for production users.
  • Commission clearer user-facing documentation explaining provider options and configuration to reduce support overhead.

View project on GitLab: ai_translate

References

[2] Add a Moderation Guardrail plugin (configurable moderation provider/model as a guardrail) (ai, Issue)

[3] Improve ai_ckeditor UX when no provider modules are installed (ai, Issue)

[8] Support long running chunking in 1.x branch for AI Search 1.x readiness (ai, Issue)

[11] Fiber branch in OpenAiBasedProviderClientBase::chat() drops token usage from the reconstructed ChatOutput (ai, Issue)

[14] Taxonomy automator clears valid term values when the LLM returns HTML-encoded entities (e.g. &) (ai, Issue)

[20] Resolve #3586531 "Add a Moderation Guardrail plugin (configurable moderation provider/model as a guardrail)" (ai, MR)

[21] feat(ai_ckeditor): warn users when no AI provider is installed in the CKEditor settings tab (ai, MR)

[34] Non-BC support for long running chunks (ai, MR)

[44] Add AI ECA Interceptor submodule: intercept AI requests/responses through ECA (ai_integration_eca, Issue)

[52] UI changes for detection of conflicts caused by external updates to underlying entity of auto-save entry (canvas, Issue)

[55] Symmetric translation: component instance version updates must propagate to all translations (canvas, Issue)

[57] Canvas AI - fix deprecated Image import, unsupported package handling, prop/slot name collision, hardcoded error messages, and add slots support (canvas, Issue)

[58] Prevention of publishing on Conflict in Canvas UI (canvas, Issue)

[60] Include language-aware indexing for ContentTemplate-rendered Canvas content (canvas, Issue)

[66] Canvas AI: add layout fixture support so page builder and template builder agent tests can run against a realistic layout (canvas, Issue)

[67] AI continues to use unsupported React libraries while creating a component (canvas, Issue)

[68] Issues/3591591: Add fixture support for layout to use in tests (canvas, MR)

[77] feat(Conflict detection): #3587526 "Detect conflicts during publishing of changes to Page entity" (canvas, MR)

[79] Add conflict detection and resolution UI for selective publishing (canvas, MR)

[80] Draft: feat: #3591596 ALTERNATE Propagate component instance updates to content/config translations; introduce `StagedLanguageConfigOverride` to treat both similarly (canvas, MR)

[82] chore(Data model): #3591679 Use string representations of expressions in... (canvas, MR)

[84] Issue #3584136: Fix canvas_component_agent prompt bugs (Image import,... (canvas, MR)

[86] #3591695 Include language-aware indexing for ContentTemplate-rendered Canvas content (canvas, MR)

[87] feat(Component sources): #3591596 Symmetric translation: component instance version updates must propagate to all translations" (canvas, MR)

[89] refactor(Data model): #3591680 Make `Coalescer` operate on `EntityFieldBasedPropExpressionInterface` objects instead of strings (canvas, MR)

[91] Draft: feat(conflict): Add conflict resolution UI with viewport selection and route integration (canvas, MR)

[94] Add context settings validation: max_global_items must not exceed max_items (ai_context, Issue)

[97] Copy and layout updates on context item scope form (ai_context, Issue)

[98] Create list of modules and configuration steps for how to integrate CCC with Document Loader MDXEditor submodule (ai_context, Issue)

[99] Add Configure Link in Modules Page (ai_context, Issue)

[102] Create CCC video showing how to configure it (ai_context, Issue)

[105] Recruit CCC beta testers (ai_context, Issue)

[115] Complete scope plugin persistence API: manager orchestration, target entity encapsulation, and plugin-driven admin/selection (ai_context, Issue)

[116] Create multi-vocabulary taxonomy scope plugin (ai_context, Issue)

[118] Issue #3586284: Clarify context item scope form copy and layout. (ai_context, MR)

[120] Issue #3586272: Add validation that max_global_items cannot exceed max_items (ai_context, MR)

[122] Issue #3586285: Add configure links to Context Control Center. (ai_context, MR)

[123] Add FilterableRagTool to augment RagTool supporting dynamic metadata filtering via additional context definitions (ai_search, Issue)

[124] How to Apply Filters in VDB for RAG Chatbot (ai_search, Issue)

[126] Attempt long running chunk indexing for 1.x branch without BC (ai_search, Issue)

[129] Issue #3584010: Add FilterableRagTool with per-index derivatives and attribute filtering (ai_search, MR)

[130] #3584014: Resolve "Attempt long running chunk indexing for 1.x branch without BC" (ai_search, MR)

[137] HTML markup gets added after certain tags (ai_translate, Issue)

[138] HTML markup is corrupted/not preserved when translating formatted text fields (text_with_summary, text_long) (ai_translate, Issue)

Period: 22 June 2026 to 23 June 2026  |  Generated: 23 June 2026 05:00 GMT

To view the full AI prompts used to generate this digest check this file: 24h_2026-06-22_prompts.txt

Looking for older digests? Browse all digests here.