Zest turns what you and your coding agents did today into a three-line standup, with the numbers behind it. It works with Claude Code, Codex, Cursor, OpenCode, Hermes and Grok Build, and with GitHub on its own.
Open the app and you see your day: three rings, your standup, your level and your team. The standup is ready at 6pm. Press it for the whole day, edit it, or share it to Slack.
A standup our own agents wrote:
🤖 31 sessions overnight, 22 PRs merged across 9 repos.
✅ Decided alone: paused 2 zero-result ad groups.
✋ Needs you: EU bid cut, Stripe key roll.
HOW IT GETS YOUR DATA
Run one command on your computer: npx @winding-labs/zest. It finds your coding agents and signs them all in with one approval. Connect GitHub and your merged PRs and commits make a standup from day one, before any agent is connected.
FIVE LEVELS
L0 Hand Coder: you write the code.
L1 Pair Programmer: the agent edits and runs things while you watch.
L2 Delegator: it plans first, tests itself, and you review the PR.
L3 Orchestrator: several agents at once, and agents review agents.
L4 Code Factory: agents start the work, ship it and report back.
Each step up is a yes/no test on your last 14 days of sessions and GitHub, never a model's opinion. The app always shows the next step as one sentence, like "2 more days with the agent running tests". The median developer on Zest runs about 3 agents a day, one at a time.
ALSO IN THE APP
• Team: a leaderboard (AI time, PRs a week, how you plan and build, the tools behind each number) and one timeline for standups, PRs and notes your agents post. Teams are worked out from who commits to which repos, so nobody sets them up.
• Spend: how much of Claude Max, Codex and GitHub Actions you've used, and what that usage would cost at API list prices.
• Ask Zest: tap the lemon and ask about your own work: "Which repo used the most CI minutes this week?" Pin the answer to Home.
• Bench: how often code review, security, performance, SEO, dependency and accessibility checks ran in the last 30 days, and which repos are overdue.
• Stack: a score out of 100 for how well you're set up (agents, models, MCP tools, skills, practices) and the gaps that would raise it.
WHY WE BUILT IT
Zest's founder went from 11 agents a day to 52 between June and November while his own AI time stayed at 7–8 hours a day. His pull requests went from 144 a week to about 980, and the median time from open to merge on those self-merged PRs fell from 21 minutes to 2. No single task got faster. Agents checking their own work, fanning out across repos and starting work on their own did it. Zest is how he saw which step mattered.
PRIVACY
The plugin removes secrets on your computer before anything is sent. Raw session logs are deleted after 30 days. Network benchmarks are anonymous and only appear when at least 10 workspaces are in the group. Your data isn't sold or used for ads.
Zest needs a Zest account and at least one connected agent or a GitHub account. It's made by Winding Labs and isn't affiliated with the makers of the tools it works with.
Questions: support@winding.ai
A 3-line standup from your coding agents' day, plus your level and your team