Refactors HistoricAiRoiStat to consolidate AI/Non‑AI group payloads into cohort_group_json, adds data_point_tooltip and contributing_committer_count, and surfaces interval_name. Updates compare-stat.js to import tooltip support and show clearer averaged labels (e.g. “Average X per week”) and hover text. Also fixes incomplete parsing of GitHub issues for Oreps. Outcome: clearer labels, richer hover info, and more accurate cohort counts.

Adds Desktop::LaunchCatchUp tests (vendor/gems/desktop_edition/test/lib/desktop/launch_catch_up_test.rb) and setup to exercise launch-time repo catch-up. Covers ordering by newest commit, skipping repos outside ACTIVITY_WINDOW (and first-launch lookback), honoring cached plans to avoid rereads, and tracking detected vs processed commit counts and completion state. Improves reliability and progress accuracy on app launch.

Add a desktop-only Job Queue UI by mounting GoodJob::Engine at /good_job when Environment.desktop? (config/routes.rb), implementing JOB_QUEUE_PATH/TITLE, jobQueueWindow and createJobQueueWindow in desktop/src/main.ts, and adding an integration test (vendor/gems/.../desktop_job_queue_dashboard_test.rb). Desktop users can now open and manage background jobs; route remains unavailable outside the desktop environment.

Adds a v1 "Diff Delta per composite token consumed" metric to AI Usage Stats and minifies claude-telemetry.mjs. TelemetryPackagesController now prefers the .min.mjs, injects a per-download api_token into the served script via a sentinel regex that survives minification, and prepends a version/build/masked-token banner. Renames token param and improves deploy artifact handling, improving usage accuracy and token traceability.

Update script/claude/claude-telemetry.mjs to improve telemetry error capture and diagnostics: add logCaughtException (writes full stack + runtime context to error.log and a sync-log pointer), build failureContext with redacted headers and a request summary, use AbortSignal.timeout for upload timeouts, unwrap fetch error causes, and bump version to v1.2. Result: clearer, safer failure logging and easier debugging of telemetry uploads.

Introduce ControllerConcerns::FileChangeSizesConcern (app/lib/controller_concerns/file_change_sizes_concern.rb) and wire it into Repos::CommitGroupsController and CommitsController to compute file_change_sizes from commit code files. Centralizes sizing logic to power a visualization of changeset directories by magnitude. Deployed to production on 2026-07-21; improves directory-level change metrics and visualization accuracy.

Added new AI usage metrics in the AI usage stats view (app/views/shared/resource_actions/ai_usage_stats.html.haml): AI code line percent, AI inactive counts by provider, Diff Delta per LLM, and comprehensive token counts (cached/uncached, input/write). Deployed to production 2026-07-16. Provides better visibility into model usage and token costs for monitoring and cost analysis.

Update SubscriptionConcerns::SeatGeneration.ensure_optimal_seat_allocation to use each committer's total commit count as a fallback when the number of seats is greater than active committers. Change implemented in vendor/gems/subscribe_manager/app/lib/subscription_concerns/seat_generation.rb to avoid under-allocation and produce a more balanced seat distribution.

Adjusts the comment impact weight in app/lib/defines/code_lines.rb: the comment impact value used during the Prompt-to-Production era was reduced from 1 to 0.1. This lowers the influence of comments in code scoring/metrics so other signals carry more weight, improving relevance of evaluations.

Implements v1 of the Maintainability Gap research feature. Adds extensive integration test coverage in test/integration/proto_user_signup_test.rb and updates config/routes.rb to expose new endpoints. Changes exercise proto user signup flows to collect AI code-quality maintainability metrics; deployed to production 2026-06-24. Benefit: enables collection and evaluation of maintainability-gap data for the research.

Introduce legacy_refactor_percent and wire it into stat calculation and segment definitions (app/lib/presenters/concerns/cached_stat_graph_value_calculate_concern.rb and app/lib/defines/stat_segments.rb). Deployed to production 2026-06-22. Allows customers to evaluate their refactor performance alongside upcoming industry benchmarks.

Add AI-Assisted Commit percentage and total token count to the default AI Usage stats. Updated the ai_usage_stats view partial to surface these new metrics so teams can see the proportion of AI-assisted commits and total tokens consumed. Released to production on 2026-06-22 20:04 — improves usage visibility and token accounting for reporting and billing.

Refresh pricing page messaging and visuals: change copy to emphasize AI ROI, update comparison/research links, remove the annual-contract line, replace the desktop hero with new light/dark hero images, and tweak introduction SCSS (gradients, alignment, spacing) for better responsive display. Key files: app/views/pages/pricing.html.haml, app/assets/stylesheets/pages/pricing/_introduction.scss, plus new hero assets. Deployed to production; improves clarity and light/dark visuals.

Update the CreateTeamMessengerNotifications job to combine notifications of the same type into consolidated messenger blocks before sending to Slack/Teams. Adds combined_messenger_blocks and best_image_block_for, and updates perform/message_channel logic to build grouped payloads. Result: fewer duplicate posts and clearer, richer grouped notifications in Slack and Teams.