Your first brainstorm
In this lesson you run /c-brainstorm for the first time. Two things happen: Cadence sets up its config for the repo, and then it converges your idea into a small written stub: the first artifact of the pipeline.
What brainstorm is for
Section titled “What brainstorm is for”Brainstorm’s job is to turn a fuzzy thought into a written starting point through disciplined question-and-answer. It asks one question at a time, each with a recommended option, and you mostly just pick. You are never asked to draft anything from a blank page. You nod or redirect.
The feature you’ll brainstorm throughout this quickstart is deliberately tiny: a function that prints the current time in a standard format. It’s small on purpose, so the loop is fast and you can compare your result against the finished example at the end.
First run: Cadence scaffolds its config
Section titled “First run: Cadence scaffolds its config”The very first time you run /c-brainstorm in a repo that has no Cadence config, Cadence notices and offers to create one.
/c-brainstorm a function that prints the current timeNo .cadence/config.yaml found in this repo or any parent.Scaffold defaults to get started? (y/n)Answer y, then answer three questions: where designs and plans should live (default docs/designs), whether plans should be test-first by default, and any helper agents to register (or “none”).
Then: the question-at-a-time loop
Section titled “Then: the question-at-a-time loop”With the config written, Cadence enters the Q&A loop. Each question leads with a plain-English sentence explaining what’s being decided and why it matters right now, gives you options, and marks exactly one as Recommended so you always have a safe default.
Brainstorm the current-time function
-
Run the command with your idea
/c-brainstorm a function that prints the current time. On a fresh repo, accept the config scaffold first (three quick questions). -
Answer one question at a time
Cadence asks focused questions: output format, whether to include a way to run it directly, what’s explicitly not in scope. Each has a Recommended pick. You choose; you don’t write prose. -
Pick an approach
Once the questions converge, Cadence proposes an approach and asks you to confirm. For this feature: a single small module with one function returning an ISO-8601 timestamp, plus a way to run it directly. -
It writes the stub
Cadence asks “I have enough — writing 00-overview.md. Proceed?” On yes, it writes a00-overview.mdstub underdocs/designs/2026-01-15-hello-cadence/(your date will differ) and stops. It does not auto-start the design stage.
The artifact: a 00-overview.md stub
Section titled “The artifact: a 00-overview.md stub”The stub is the artifact you’ll read. It’s short and plain-English: a “What we’re building” paragraph, a one-line “Why”, the chosen approach, a proposed list of child documents, and a decisions log capturing every choice you made during the Q&A. You can open the finished version of exactly this stub in the worked example:
docs/designs/2026-01-15-hello-cadence/00-overview.md- What we’re building: a
current_time_iso()function returning an ISO-8601 string. - Why: the canonical Cadence teaching example.
- Approach: one module, one function, one test, run-directly entry.
- Doc index: the child docs design will write next.
- Decisions log: every Recommended pick you accepted.
Next: Design and plan it. Grow the stub into a full design, then translate it into an exact plan.