Omakase-Off

Chef's choice exploration - when you're not sure WHAT to build, explore different approaches in parallel.

Part of Test Kitchen Development:

• omakase-off - Chef's choice exploration (different approaches/plans)
• cookoff - Same recipe, multiple cooks compete (same plan, multiple implementations)

Core principle: Let indecision emerge naturally during brainstorming, then implement multiple approaches in parallel to let real code + tests determine the best solution.

Three Triggers

Trigger 1: BEFORE Brainstorming

When: "I want to build...", "Create a...", "Implement...", "Add a feature..."

Present:

Before we brainstorm the details, would you like to:

1. Brainstorm together - We'll explore requirements and design step by step
2. Omakase (chef's choice) - I'll generate 3-5 best approaches, implement them
   in parallel, and let tests pick the winner

Trigger 2: DURING Brainstorming (Indecision Detection)

Detection signals:

• 2+ uncertain responses in a row on architectural decisions
• Phrases: "not sure", "don't know", "either works", "you pick", "no preference"

When detected:

You seem flexible on the approach. Would you like to:
1. I'll pick what seems best and continue brainstorming
2. Explore multiple approaches in parallel (omakase-off)

Trigger 3: Explicitly Requested

• "try both approaches", "explore both", "omakase"
• "implement both variants", "let's see which is better"

Workflow Overview

| Phase | Description | |-------|-------------| | 0. Entry | Present brainstorm vs omakase choice | | 1. Brainstorm | Passive slot detection during design | | 1.5. Decision | If slots detected, offer parallel exploration | | 2. Plan | Generate implementation plan per variant | | 3. Implement | Dispatch ALL agents in SINGLE message | | 4. Evaluate | Scenario tests → fresh-eyes → judge survivors | | 5. Complete | Finish winner, cleanup losers |

See references/detailed-workflow.md for full phase details.

Directory Structure

docs/plans/<feature>/
  design.md                  # Shared context from brainstorming
  omakase/
    variant-<slug>/
      plan.md                # Implementation plan for this variant
    result.md                # Final report

.worktrees/
  variant-<slug>/            # Omakase variant worktree

Slot Classification

| Type | Examples | Worth exploring? | |------|----------|------------------| | Architectural | Storage engine, framework, auth method | Yes | | Trivial | File location, naming, config format | No |

Only architectural decisions become slots for parallel exploration.

Variant Limits

Max 5-6 implementations. Don't do full combinatorial explosion:

1. Identify the primary axis (biggest architectural impact)
2. Create variants along that axis
3. Fill secondary slots with natural pairings

Critical Rules

1. Dispatch ALL variants in SINGLE message - Multiple Task tools, one message
2. MUST use scenario-testing - Not manual verification
3. Fresh-eyes on survivors - Required before judge comparison
4. Always cleanup losers - Remove worktrees and branches
5. Write result.md - Document what was tried and why winner won

Skills Orchestrated

| Dependency | Usage | |------------|-------| | brainstorming | Modified flow with passive slot detection | | writing-plans | Generate implementation plan per variant | | git-worktrees | Create isolated worktree per variant | | parallel-agents | Dispatch all variant subagents in parallel | | scenario-testing | Run same scenarios against all variants | | fresh-eyes | Quality review on survivors → input for judge | | finish-branch | Handle winner (merge/PR), cleanup losers |

Example Flow

User: "I need to build a CLI todo app."

Claude: [Triggers omakase-off]
Before we dive in, how would you like to approach this?
1. Brainstorm together
2. Omakase (chef's choice)

User: "1"

Claude: [Brainstorming proceeds, detects indecision on storage]

You seem flexible on storage (JSON vs SQLite). Would you like to:
1. Explore in parallel - I'll implement both variants
2. Best guess - I'll pick JSON (simpler)

User: "1"

[Creates plans for variant-json, variant-sqlite]
[Dispatches parallel agents in SINGLE message]
[Runs scenario tests on both]
[Fresh-eyes review on survivors]
[Presents comparison, user picks winner]
[Cleans up loser, finishes winner branch]