AI Scrum Master · in your terminal
Best friend to engineers and agents.
yeaboi turns a project description into epics, stories, tasks, and a sprint plan — then runs the standups, retros, poker, 1:1s, and delivery reports around them. And because half your team is AI now, it watches those agents too: what they cost, what they shipped, and what they're allowed to do. No browser tabs. No ceremony overhead. Just your terminal.
uv tool install yeaboi
or pipx install yeaboi · Python 3.11+ · MIT
00 / two worlds
Who are we working with today?
Your team is not all human any more. yeaboi opens on the question: the people running the sprint, or the agents writing the code. Pick a side and the whole app follows it.
01 / analysis
Know how your team actually delivers.
Reads your Jira or Azure DevOps history — velocity, estimation accuracy, churn — and builds a team profile that calibrates every plan yeaboi makes after it.
02 / planning
Description in, sprint plan out.
A five-stage pipeline — analysis, features, stories, tasks, sprints — with a review checkpoint after every stage. Accept, edit, or reject; then push to your board.
03 / standup
The daily that runs itself.
Detects what everyone shipped from commits and tickets, scores sprint-day confidence, and delivers the summary to terminal, Slack, or email — on an OS schedule, even with yeaboi closed.
04 / retro
A board the whole team joins.
Spins up a live board any browser on the network can join — cards, reactions, a shared timer. When the talking's done, AI drafts the action items.
05 / poker
Estimate together, from the terminal.
Opens a live voting board the team joins from any browser. Everyone votes, AI reads the spread and flags where you disagree, and the agreed points sync straight back to your board.
06 / performance
1:1s backed by real work.
Preps each 1:1 from the engineer's actual delivery and last meeting's open actions, writes up the summary after, and synthesizes six-month reviews when they're due.
07 / reporting
Tell the business what shipped.
Turns a sprint, a month, or a whole quarter of delivered work into an executive summary — as a slide deck, HTML, or Markdown.
08 / usage
Costs, in the open.
Token usage, session history, and cost estimates for every LLM call — tracked per session, right in the app.
09 / settings
Configured without leaving the terminal.
API keys, LLM provider, board connections, log level, data directory — all managed in-app, validated as you go.
10 / agent usage
Find out what your agents cost.
Reads the session logs your coding agents already write and totals them: tokens, cache hits, spend per model and per project, and the daily trend. Computed on your machine — nothing is uploaded.
11 / agent standup
A standup for the machines.
The same daily digest you get for the team, for the agents working beside it: which projects they worked, what landed as commits and PRs, and which threads are still open.
12 / agent security
Know what they're allowed to do.
Audits the setup behind your agents — permission rules, MCP servers, secrets sitting where a session can read them, risky commands already run — and scores the posture.
plugs into your stack
$ yeaboi — pick a mode
Two worlds. One command.
Everything a lead does between planning and delivery, each its own screen in the TUI. Read-only sources plug in when you have them; skip anything you don't — and every generated step pauses for your review, so nothing lands without your [1] Accept.
humans — run your team's scrum
Planning
Turn a project description into epics, stories, tasks, and a sprint plan — then push straight to Jira or Azure DevOps.
jira · azure devopsAnalysis
Read your team's real velocity and history first, so estimates and capacity are calibrated to how you actually deliver.
velocity · estimation · churnStandup
Detect what everyone shipped, score sprint progress, and deliver the summary — on a schedule, even when yeaboi is closed.
scheduled · slack · emailRetro
Open a board your whole team joins from any browser, anywhere. Cards, reactions, a shared timer — then AI action items.
live board · secure tunnelPoker
Run planning poker from the terminal — the team votes in a browser, AI reads the spread, and the points sync back to your board.
live voting · points syncPerformance
BetaPrep and write up 1:1s, and synthesize six-month reviews from each engineer's real delivery history. Beta — output is not yet verified against real delivery data.
1:1s · six-month reviewsReporting
Turn a sprint or a whole quarter of delivered work into a business-ready summary and a slide deck for stakeholders.
slides · html · markdownagents — watch your AI agents work
The other half of the team doesn't come to standup. yeaboi reads the session logs your coding agents already write — on your machine, never uploaded — and answers the three questions nobody else is asking about them.
Agent Usage
BetaWhat your agents cost: tokens, cache hits, spend per model and per project, and the daily trend.
tokens · cache · spendAgent Standup
BetaA daily digest of what they did — sessions worked, commits and PRs landed, and the threads still open.
daily digest · commits · prsAgent Security
BetaAudit the setup behind them: permission rules, MCP servers, secrets exposure, and risky commands already run.
permissions · mcp · secretsthree lines to your first plan
Plan your next sprint tonight.
Installs in its own isolated environment. Works with Anthropic Claude by default — plus OpenAI, Google, and AWS Bedrock.
uv tool install yeaboi
1 install
uv tool install yeaboi
or pipx install yeaboi
2 configure
yeaboi --setup
adds your API key
3 run
yeaboi
launches the interactive TUI
Optional extras: uv tool install "yeaboi[voice]" for offline voice input, or "yeaboi[all-providers]" for every LLM provider.
yeaboi also ships as an MCP server and a Claude Code plugin — run every mode straight from Claude Code, Cursor, or any MCP client.
