new-idea

/new-idea — Brainstorm, Research & Document Product Ideas

Captures product ideas through structured brainstorming, market research, and codebase analysis. Produces a living document in the ideas directory that can be directly ported into a Ralph sprint via /new-sprint.

Before Starting

Read RALPH.md to find:

• Documents.ideas — path to the ideas index file (derive the ideas directory from this path)
• Documents.prd — path to the PRD

If RALPH.md doesn't exist or is missing these paths, tell the user: "RALPH.md is missing or incomplete. Run /ralph-init to set up your project."

Step 1: Explore the Idea

Start a conversation to understand what the user is thinking. Don't jump to solutions — dig into the problem first.

Ask about (naturally, not as a checklist):

• The problem: What user pain point does this solve? What's frustrating today?
• The vision: What would it look like when it's working? Walk me through the experience.
• Who benefits: Which user persona cares most? (See your PRD for user personas)
• Inspiration: Did they see this somewhere? What triggered the idea?

Keep going until you have a clear picture. Summarize back to the user: "So the idea is..." and confirm.

Step 2: Research the Market

Use WebSearch to investigate how others handle this. Research at least 3 angles:

1. Direct competitors: Other apps in this space that have this feature — how do they implement it? What's good/bad?
2. Adjacent apps: Apps outside this space that solve a similar UX problem — what patterns work?
3. Technical approaches: How is this typically built? What libraries, APIs, or services exist?

Present findings as a brief competitive landscape:

## Market Research

### How Others Do It
- **[App/Service]**: [How they handle it, what works, what doesn't]
- **[App/Service]**: [...]

### What We Can Do Better
- [Specific differentiator or improvement opportunity]

### Technical Options
- [Approach 1]: [trade-offs]
- [Approach 2]: [trade-offs]

Discuss findings with the user — do they want to go a similar direction or differentiate?

Step 3: Audit the Codebase

Silently check what already exists that's relevant:

1. Read CLAUDE.md — current tech stack, project structure (if exists)
2. Read AGENTS.md — codebase conventions, key files (if exists)
3. Read README.md — project overview (if exists)
4. Read the PRD (from RALPH.md Documents.prd) — check if this idea overlaps with planned features
5. Search the codebase for related files, hooks, components, or database tables
6. Read the ideas index (from RALPH.md Documents.ideas) — make sure this isn't a duplicate of an existing idea

Present a brief summary:

## Codebase Relevance

### Already Exists
- [What we'd build on — existing tables, hooks, components, APIs]

### Would Need
- [New tables, services, screens, or integrations required]

### PRD Overlap
- [Does this overlap with any planned PRD features? Which section?]

Step 4: Assess the Idea

Rate the idea on these dimensions. Be honest — the user wants real assessment, not cheerleading.

## Assessment

| Dimension | Rating | Notes |
|-----------|--------|-------|
| **User Value** | High / Medium / Low | How much does this improve the user experience? |
| **Effort** | Trivial / Easy / Medium / Hard / Very Hard | How complex is this to build? |
| **Technical Risk** | High / Medium / Low | Unknown APIs, complex state, performance concerns |
| **Scope Creep Risk** | High / Medium / Low | How likely is this to balloon beyond the original idea? |
| **Market Fit** | Strong / Moderate / Weak | Does this differentiate us or just match competitors? |

**Overall**: [1-2 sentence honest take — is this worth pursuing and when?]

Effort Scale

• Trivial: A few hours, minor changes to existing code
• Easy: 1 sprint (3-5 stories), no new infrastructure
• Medium: 1-2 sprints, may need a new API or service
• Hard: 2-3 sprints, new infrastructure or major refactor
• Very Hard: 3+ sprints, foundational change or new platform capabilities

Risk Flags

• Depends on external API with no fallback
• Requires ML/AI model training or fine-tuning
• Changes core data model (migrations across many tables)
• Needs real-time sync or complex state management
• Platform-specific behavior divergence

Discuss the assessment with the user. They may want to adjust scope to reduce effort or risk.

Step 5: Document the Idea

Derive the ideas directory from the Documents.ideas path in RALPH.md (e.g., if ideas is docs/ideas/_index.md, the directory is docs/ideas/). Write the idea file to [ideas directory]/[idea-name].md using the template below. The format is designed to port directly into /new-sprint when the time comes.

File Naming

• Use kebab-case: smart-search.md, bulk-import.md
• Keep names short and descriptive

Template

# [Idea Title]

> [One-line summary of what this does for the user]

**Status:** idea
**Proposed by:** [user]
**Date:** [YYYY-MM-DD]
**Effort:** [Trivial/Easy/Medium/Hard/Very Hard]
**Value:** [High/Medium/Low]

## Problem

[What user pain point does this solve? Why does it matter?]

## Proposed Solution

[High-level description of the feature. What does the user experience look like?]

## Market Research

### Existing Solutions
- **[App/Service]**: [How they handle it — strengths and weaknesses]

### Our Differentiator
[What would make our version better or different?]

### Technical Options
- [Approach]: [trade-offs]

## Technical Considerations

### Builds On
- [Existing tables, hooks, components, APIs we'd leverage]

### New Infrastructure
- [New tables, services, screens, or integrations needed]

### Risks
- [Technical risks, scope creep risks, dependency risks]

## Open Questions

- [Unresolved questions that need answers before implementation]

## PRD Mapping

[Which PRD section would this fall under? Is it an extension of an existing feature or entirely new? This helps `/new-sprint` know where to slot it.]

Status Values

• idea — just captured, not yet committed to
• researched — has market research and technical audit
• planned — committed to building, ready for /new-sprint
• in-progress — currently in a sprint
• shipped — built and merged to main
• parked — intentionally deferred (add reason)

Step 6: Update the Index

Add or update the entry in the ideas index file (from RALPH.md Documents.ideas).

If the index file doesn't exist, create it from the template:

# Ideas

Living index of product ideas. Use `/new-idea` to add new entries.

| Idea | Status | Effort | Value | File |
|------|--------|--------|-------|------|

The index format:

| [Title] | idea | Medium | High | [filename.md](filename.md) |

Sort by status: planned first, then researched, then idea, then parked. Within each group, sort by value (High first).

After Documenting

Tell the user the idea has been saved, then ask what they'd like to do next:

Use AskUserQuestion:

• Promote to backlog now — "The idea feels fully baked. Let's resolve open questions and add it to the backlog." If selected, invoke /promote-idea for this idea (skip Step 1 of that skill since the idea is already identified).
• Save for later — "I want to think on it. I'll run /promote-idea when I'm ready."
• Brainstorm another idea — Loop back to Step 1.

Idea documented in [ideas directory]/[filename].md

When you're ready to build it, run /promote-idea to move it
to the backlog, then /new-sprint to generate a sprint.