ideate

name	ideate
version	2.10.2
description	Use when the user has an idea or requirement to decompose into backlog stories.
trigger	manual
depends-on	[]
references	["references/story-template.md"]
disable-model-invocation	true
user-invocable	true
allowed-tools	Read, Glob, Grep, Bash, Edit, Write, WebSearch, WebFetch, Agent
argument-hint	<idea-or-requirement>

---

ideate

Skill metrics: Emit a start event to the activity log:

echo "{\"type\":\"skill\",\"event\":\"start\",\"skill\":\"ideate\",\"ts\":\"$(date -u +%Y-%m-%dT%H:%M:%SZ)\"}" >> docs/sessions/.activity-log.jsonl

Transforming idea into backlog stories: $ARGUMENTS

Step 0: Check Discovery State

Discovery has been completed. Load `docs/reference/DECISION_LOG.md` and `docs/reference/ASSUMPTION_REGISTER.md` for story generation context. When generating stories, include relevant entries in each story under "Relevant Decisions", "Relevant Assumptions", and "No-Gos" sections. Discovery hasn't been completed. Warn the user: > "Discovery hasn't been completed. Run `/discover` first for thorough guided discovery, or proceed with `/ideate` for quick decomposition (less guidance, more assumptions)." Wait for user decision before proceeding. No vision files found — proceed with current behavior (no change).

Phase 0: Validate Prerequisites

Before starting, verify:

• docs/reference/backlog/ directory exists (create if missing)
• docs/reference/BACKLOG_INDEX.md is readable
• No conflicting ideation in progress (check for uncommitted backlog changes)

If prerequisites fail, inform the user and stop — don't consume context on doomed work.

Sprint Scope Awareness

Check if a sprint is currently active (branch matches sprint-* or docs/progress.md shows an in-progress sprint). If so, display a notice:

Sprint [N] is in progress (goal: "[sprint goal]"). New stories will be added to the backlog, not the current sprint. Adding stories to a running sprint increases sprint churn — target is <20%.

This is informational, not blocking. The user may legitimately need to ideate during a sprint. But it prevents accidental scope creep where new stories get silently mixed into active sprint work.

1. Gather Input

Check for prior brainstorm artifacts: Search docs/brainstorms/ for existing design documents matching the topic. If a brainstorm doc exists with status: decided, load it as context — the design exploration and key decisions have already been made.

Understand the idea or requirement from $ARGUMENTS and any additional context the user provides.

Ask clarifying questions if the idea is too vague:

Do NOT write any stories to backlog files, create epic files, or update BACKLOG_INDEX.md until the user has explicitly approved the decomposition. Present the stories, wait for approval.