**Date:** Monday, June 29, 2026 **Attendees:** Aidan, Marcus, Ángela, Darren, Emma ### Canvas chat ticket * Stop becomes play/pause; resume re-prompts the agent to continue, possible because generation runs one loop per request * Stop and undo is renamed Stop and revert all changes, a milestone revert of the whole turn, kept redo-able and distinct from native Ctrl-Z ### AI Content Review criteria-creator UI * A form that writes the agent's markdown prompt: guidelines, per-criterion thresholds, and scored good/medium/bad examples * Two scoring methods get different UIs: example-based, and a count-the-bads penalty for do/do-not guidelines * Deterministic checks (SEO, Flesch-Kincaid, color, banned words) use tools; the agent explains the result and what to fix * Default is a small text-only agent; advanced mode adds a tool-using agent; scoring and feedback stay separate agents * Feedback agent needs an annotation tool to place markers on the page; AI Content Review scopes to editor-fixable issues, with developer items escalated ### Modules to evaluate * AI Metering: adopt and recommend the existing WebShake usage module rather than build one * Annotation: role-scoped field annotations including AI-agent-only notes; a missing CCC piece, pending UX and accessibility review ## Decisions * Canvas Stop becomes play/pause; Stop and revert all changes does a redo-able milestone revert * The criteria UI writes the agent markdown prompt; default text-only agent, advanced mode adds tools * Scoring and feedback are separate agents; the feedback agent uses a dedicated annotation tool * Adopt the WebShake metering module and the Annotation module rather than rebuilding ## Questions * Should the criteria UI move back toward standard Drupal form patterns, and how do we shape the count-the-bads scoring UI? * Where does editor-fixable end and developer escalation begin for review feedback? ## Next Steps * **Aidan:** Finalize the Canvas chat ticket wording; write functional requirements for the feedback/annotation agent and criteria UI * **Ángela:** Review and demo AI Metering (priority) and Annotation (secondary), with improvement ideas * **Marcus:** Continue adopting the metering module; offer a meeting with the Annotation author