Discount with code applied — 50% off Pro or Max for 6 months
AI CODING ANALYTICS
In Claude Code, /usage shows your plan usage and session cost, and Team and Enterprise admins get a dashboard of users, sessions and accepted lines. Zest adds the team view: each engineer's sessions, models, tokens and skills next to the pull requests they merged.
Free for 3 seats · No credit card
01See your team’s AI stack
Winding / Teams Last 30 days 
Zest
| Member | AI | PRs/wk | AI Task | Plan | Impl | Last Standup | AI Stack |
|---|---|---|---|---|---|---|---|
| 22.8/wk | 28 min ago | ||||||
| 15.8/wk | 12 min ago | ||||||
| 4/wk | Aug 14 | ||||||
| — | 2 hours ago |
Demo setups use current model families. The pairings above are the combinations benchmarked by Artificial Analysis; rankings measure performance, not popularity.
Illustrative daily trends · Other totals: last 30 days. Build Hours: this week, model-estimated activity including overlap; not engineer hours.
USED BY ENGINEERING TEAMS AT
For one person, Claude Code answers inside the terminal:
ccusage, an open-source CLI, reads the agent logs on one machine (Claude Code, Codex and a dozen other CLIs) into daily, monthly, session and 5-hour-block reports, and has a status line mode. Claude Code's /stats tab reads local history too. Both stop at one laptop, and Claude Code deletes transcripts after 30 days by default.
Local counts can disagree. On the Claude Code version we ran, the cache behind /stats counted 1.9 to 2.7 times the tokens of a deduplicated recount of the same transcripts (September 5 to 12, 2026), because each response's usage is logged once per content block. The status line's own cost estimate came within 8% of our recount. And one laptop isn't the team: in the 30 days to October 3, our heaviest user's laptop opened 596 of the 878 pull requests we merged (68%); the rest came from three other engineers, Codex and other machines.
Use ccusage for your own laptop. Zest is the team view: every engineer's sessions across coding agents, sent from each machine and joined to the pull requests they merged.
Each engineer installs the Zest plugin with /plugin marketplace add https://github.com/Winding-Labs/zest-claude, then /plugin install zest, then /reload-plugins, then /zest:login, which takes about a minute. From then on, three hooks record session start, session end and every tool call, and each session lands on the team dashboard.
Claude Code can export telemetry over OpenTelemetry: set CLAUDE_CODE_ENABLE_TELEMETRY=1, pick an exporter (OTLP, Prometheus or console) and point it at a collector you run. It emits counters for sessions, lines of code, commits, pull requests, tokens and cost; Anthropic notes the cost metrics are approximations.
Zest needs no collector, and it doesn't read OpenTelemetry exports, so the two run side by side.
Of the 2,620 pull requests merged in our two repos from July 5 to October 3, 2026, 95% carry an AI agent's signature and about 90% carry Claude Code's.
866 pull requests merged in our two repos from August 12 to October 4, 2026 carry a session-analytics block, written by the agent that opened them. The blocks show the default model changing twice in four weeks: Fable 5 and Opus 5 in August, then Fable 5.1 from September 1, then Opus 5.5 from September 22, which wrote every pull request in October. Opus 5.5's merged in a median 25 minutes and Opus 5's in 47, over different weeks and different work.
The Zest GitHub App matches each Claude Code session to the pull request it led to, by the branch, commits and PR links the session touched.
TRY IT ON YOUR TEAM
Starter is free (3 seats, no credit card). Pro is $49 a month with 5 seats and Max is $149 a month with 10 seats — priced per workspace, not per engineer, with extra seats at $19 a month.
A FEW THINGS TO KNOW
Install the Zest plugin in each engineer's Claude Code (/plugin marketplace add https://github.com/Winding-Labs/zest-claude, then /plugin install zest, then /reload-plugins, then /zest:login) and connect your GitHub org. Sessions appear on the Zest dashboard per engineer, with their models, tokens, skills and the pull requests they led to.
No code diffs are uploaded. The Claude Code plugin sends the text of the conversation, not tool inputs or outputs, after redacting secrets and personal data on the engineer's machine. Message text is kept for at most 90 days.
