/moonshot

Stop. Step back from the backlog, the roadmap, the current sprint. Answer one question:

What's the single smartest, most radically innovative, accretive, useful, and compelling addition you could make to this project at this point?

Rules

Exactly one answer. Not a list. Not options. One thing, fully committed.
Argue it. Why THIS over everything else? What makes it the inflection point?
Be bold. This is not incremental improvement. This is the move that changes the trajectory.
Be concrete. Name the feature, the architecture, the integration — not a vague direction.
Be honest. If the project needs something unglamorous, say that. Innovation includes knowing when the highest leverage is boring infrastructure.

Process

1. Deep Immersion

Before answering, build a complete mental model:

Read the full codebase structure (.glance.md files, READMEs, CLAUDE.md)
Read the backlog (GitHub issues, .groom/BACKLOG.md)
Read recent git history (last 30 commits minimum)
Read any project.md, retro files, or ADRs
Understand the user, the product, the market context

2. Divergent Search

Generate at least 10 candidates internally. For each, evaluate:

Leverage: How much value per unit of effort?
Innovation: Does this exist elsewhere? Is this a novel combination?
Accretion: Does this compound? Does it make future work easier/better?
Usefulness: Does this solve a real problem or create real capability?
Timing: Why now? What makes this the right moment?

3. Convergent Selection

Pick one. Kill your darlings. The answer should survive this gauntlet:

"If we could only ship one more thing, would this be it?"
"Will this still matter in 6 months?"
"Does this unlock things that are currently impossible, not just inconvenient?"
"Is this the kind of thing that makes people say 'why didn't we do this sooner'?"

4. The Pitch

Present your answer as:

## The Move

[One sentence: what it is]

## Why This, Why Now

[2-3 paragraphs: the argument. What changes. What it unlocks.
Why the timing is right. Why everything else is less important.]

## What It Looks Like

[Concrete description: architecture, UX, integration points.
Enough detail to judge feasibility.]

## What It Costs

[Honest effort estimate. What gets displaced. What risks exist.]

## The Unlock

[What becomes possible AFTER this ships that isn't possible today?]

Anti-Patterns

Listing 5 ideas and asking the user to pick (that's /groom)
Proposing something the project already does
Vague directional advice ("invest in testing", "improve DX")
Safe, incremental, obvious next steps (that's the backlog)
Ignoring constraints (team size, tech stack, market position)

Creative Reframing

When the divergent search feels stuck producing variations instead of new ideas, load references/break-the-frame.md for signal detection and hard reframing moves.

Composability

After /moonshot, the user may:

/groom — to fit the moonshot into the backlog
/shape — to plan the moonshot in detail
/research thinktank — to stress-test the idea
/autopilot — to just build it