Release Notes

Verbal v1.0.0 released

The first major milestone — Verbal v1.0.0 is officially out.

Redesigned sidebar navigation

The dashboard sidebar now has subsections, a theme toggle, and a quick-peek panel so you can jump to what you need faster.

Sessions are automatically named

Verbal now reads your session content and generates a meaningful name automatically — no more "Session 42."

Sign in with Google works seamlessly

Google OAuth is fully wired up, including automatic account creation on first sign-in.

CLI authentication no longer requires a local server

The CLI now authenticates via browser polling — simpler setup, no ports to open.

Personalized prompt coaching

Coaching tips are now tailored to your skill level, filtered by anti-patterns you actually exhibit, and rotated so you see fresh advice each visit.

Session Report Cards

Every session now gets a detailed report card showing quality scores across five dimensions, with coaching tips specific to that session.

Google OAuth sign-in loop resolved

Users signing in with Google no longer get redirected in a loop.

Claude Max plan picker restored in CLI setup

The Claude Max subscription option is now selectable again when running the CLI setup wizard.

More accurate PR and commit counts per session

PR and commit counts are now scoped to the correct session, and duplicate entries have been removed.

Dashboard charts, coaching, and intent breakdown polished

Several display issues with charts, coaching card layout, and the intent mix visualization have been cleaned up.

Better empty states and refined ROI calculations

Empty states now give you a clear next action, and the ROI formula has been tightened up.

Richer telemetry with subagent filtering

Agent traces now carry more context — external IDs, resource attributes, and subagent filtering — for more accurate usage attribution.

Email notifications match the Verbal brand

All outgoing emails have been refreshed to use the updated visual identity.

Dashboard loads faster with pre-computed stats

Event counts and token totals are now read from summary tables instead of scanning raw events on every page load.

ROI calculations are now significantly faster

The ROI page queries pre-aggregated daily stats instead of scanning the full events table.

All data shows correctly without truncation

A row-limit issue in the data layer was causing some results to be silently cut off. This is now fixed.

Light mode

Verbal now supports both dark and light themes. Toggle between them from the sidebar.

CLI imports your usage history automatically

On first run, the CLI now pulls in your existing Claude Code usage so your dashboard is populated immediately.

Weekly dashboard with radar chart and peak hours

The weekly view now includes a quality radar chart and a peak-hours breakdown so you can see when and how you use AI most effectively.

Redesigned ROI section with clearer spend breakdowns

The headline ROI card now shows a unified spend view with estimated and actual cost labels side by side.

Developer activity section on Explore and Home

See your AI-powered development patterns — skills, velocity, and cost efficiency — at a glance on the main dashboard views.

Deep-dive developer sections

New drill-down sections show skill progression and cost-per-outcome metrics for your development work.

Expanded session viewer with five tabs

Click any session to open a full-page view with Conversation, Analysis, Telemetry, Loops, and Overview tabs.

Session report cards with developer activity

Each session card now shows your coding activity alongside AI usage for a complete picture of the work done.

Click any metric to drill down

Dashboard metrics are now interactive — click to drill down, or use the peek panel to inspect sessions without leaving your current view.

Unified usage dashboard

AI usage, developer activity, skills, and telemetry are now presented in one cohesive view.

Direct integration with Anthropic Admin API

Verbal now connects to the Anthropic Admin API for accurate billing data.

Real-time session performance metrics

Token speed, duration, and performance data are now collected and displayed per session.

Completely redesigned spend page

The spend page now covers subscription tracking, API spend, and ROI analysis in one place.

Redesigned session viewer

Sessions now have a collapsible metadata sidebar and expandable thinking blocks so you can focus on what matters.

OpenTelemetry integration

Verbal now accepts OTLP logs and metrics, enabling automatic usage tracking directly from Claude Code.

Privacy level selection for tracking

Choose how much data Verbal captures — everything, usage only, or nothing at all.

Smoother session browsing experience

Eight session UX improvements shipped, including better scroll and load behavior.

Sessions now show token counts prominently

Token usage is displayed first in session cards for quicker cost scanning.

More detailed per-session performance metrics

Duration, token speed, CLI version, and tool usage stats are now collected for every session.

Dashboard reorganized into six zones

Navigation is now structured around Home, Explore, Flow, Sessions, Insights, and Manage for clearer information architecture.

Weekly and monthly summary dashboards

New time-based views give you a trend-level summary of your AI spend and quality over the week or month.

Every dollar shows estimated vs. actual

Cost Transparency labels make it clear which numbers are estimates and which come from real billing data.

More accurate and trustworthy dashboard numbers

Four credibility bugs have been fixed — the numbers you see now reflect your actual usage.

Full transparency into how costs are calculated

The Cost Provenance framework is live — every number in Verbal now has a clear explanation of where it comes from.

Import provider billing data and replay sessions

You can now import billing exports from AI providers and replay individual sessions for analysis.

More consistent cost calculations across the app

A unified cost query layer means spend numbers are now computed the same way everywhere in Verbal.

Browse sessions with expandable cards

The sessions page now uses progressive disclosure — cards show a summary by default, and expand to reveal full detail on demand.

Centralized prompt scoring engine

All prompts are now scored across five dimensions (clarity, specificity, structure, efficiency, and context) using a unified analysis engine.

Fresh visual identity with warm coral and hunter green accents

Verbal has a new brand system built on oklch color tokens — warmer, more distinctive, and designed to scale.

Consistent colors throughout the app

A full semantic color sweep ensures design tokens are applied correctly across every surface.

Refined billing comparison and automatic session naming

Billing comparisons are more accurate, and sessions now get names assigned automatically after they complete.