Quest Y-1: Ship the Stand-up Bot
Status: drafted · Time: 360 min · Audience: daily-builder Outcome: Build a small working utility with Claude Code, demo it, and explain the prompt, context, and harness decisions.
Win condition: a small utility runs, solves a real repeated task, and has a short demo plus a Three Pillars retro.
The quest name says “Stand-up Bot,” but the shape matters more than the exact utility. Build something small that saves repeated effort: a Slack formatter, a docs summarizer, a spreadsheet cleanup script, a weekly update drafter, or a stand-up digest helper.
Prerequisite
Section titled “Prerequisite”You need:
- White Belt complete;
- Yellow modules Y01-Y08 read;
- a GREEN local environment;
- one repeated task worth making lighter;
- a repo, gist, or approved workspace where the utility can live.
If the task touches sensitive data, choose another task or reduce it to redacted sample input.
The task
Section titled “The task”Build a small utility end-to-end with Claude Code in a single working day.
The utility should:
- accept an input you can safely share;
- produce a useful output;
- run more than once;
- have a short README or usage note;
- be demoable in under two minutes.
The utility does not need to be production software. It does need to run.
Four copyable ideas
Section titled “Four copyable ideas”Stand-up formatter. Input: rough bullets. Output: structured yesterday/today/blockers update.
Docs summarizer. Input: a redacted meeting note. Output: decisions, action items, open questions.
Spreadsheet cleanup helper. Input: sample CSV with fake data. Output: normalized column names or sorted rows.
PR description drafter. Input: git diff --stat plus your notes. Output: PR sections you review before using.
Pick the one closest to your real work.
Worked build loop
Section titled “Worked build loop”Prompt:
Goal: build a tiny stand-up formatter utility.Context: I often start with rough bullets and need a clean update.Input shape: plain text bullets with optional labels.Output shape: Yesterday, Today, Blockers.Constraints:- use sample data only;- keep implementation in one small file plus README;- do not add dependencies unless necessary;- show a plan before editing.Success criteria:- I can run it locally;- README shows usage;- sample input produces structured output.After plan:
Apply the smallest implementation. Then show me:1. files changed;2. how to run it;3. sample output.Then run the command yourself. Do not claim success from code alone.
Three Pillars retro
Section titled “Three Pillars retro”Your retro must name:
| Pillar | What to record |
|---|---|
| Prompt | Which prompt made the task clearer? |
| Context | What input, examples, or constraints did Claude need? |
| Harness | What command, test, README, or demo made the output reviewable? |
Example:
Prompt: the useful move was defining input and output shapes before asking for code.Context: I gave two fake stand-up examples and the sections I wanted.Harness: I kept it to one script and a sample command in README, then ran it locally.Evidence template
Section titled “Evidence template”## Quest Y-1 evidence
Builder handle:Date:Utility name:Repo or artefact URL:Demo link or screenshot:How to run:
Prompt decision:Context decision:Harness decision:
Reviewer or pair:Follow-up needed:What counts
Section titled “What counts”This counts:
- a utility that runs locally;
- a README or usage note;
- a short demo;
- a retro that names prompt, context, and harness decisions;
- safe sample input.
This does not count:
- a tool that does not run;
- a prompt-only idea with no artefact;
- a utility built without Claude Code;
- a retro that only says “Claude wrote it”;
- use of sensitive raw work data.
Common failure modes
Section titled “Common failure modes”“I picked a task too large for one day.” Cut it down to one input and one output.
“The utility works only on my exact sample.” That may be fine for Yellow if the sample is representative. Name the limitation.
“Claude added dependencies.” Ask whether the task can be done without them. Small utility first.
“The demo is confusing.” Demo one happy path. Save edge cases for later.
Success criteria
Section titled “Success criteria”You pass Quest Y-1 when:
- the utility runs;
- usage is documented;
- demo evidence exists;
- the Three Pillars retro exists;
- no sensitive material is included.
What you can say after this quest
Section titled “What you can say after this quest”“I can build a small useful utility with Claude Code and explain the decisions that made it work.”
Previous: Y.14 Staying current - Next: Quest Y-2 The 30-day challenge