Nw Rigor

NW-RIGOR: Quality Profile Selection

Wave: CROSS_WAVE | Agent: Main Instance (self) | Command: /nw-rigor [profile]

Overview

Interactive command to select a quality-vs-token-consumption profile. Persists choice to either ~/.nwave/global-config.json (global scope) or .nwave/des-config.json (project scope) under the rigor key. All wave commands read this config to adjust agent models, review policy, TDD phases, and mutation testing.

You (the main Claude instance) run this directly. No subagent delegation.

Note on TDD phase canons (dual-canon, ADR-025, 2026-05-07): the canonical TDD cycle is 3-phase v5 (RED, GREEN, COMMIT) as described in nw-tdd-methodology. The legacy 5-phase v4 (PREPARE, RED_ACCEPTANCE, RED_UNIT, GREEN, COMMIT) is preserved for backward-compat audit-log replay of pre-2026-05-07 commits. The .nwave/des-config.json tdd_phases field accepts BOTH canons: writers of new logs may emit v5; the validator + schema (step-tdd-cycle-schema.json) dispatch on schema_version ("5.0" → canonical, anything else → legacy). The profile table below shows both canons side-by-side: pick v5 for new features, keep v4 only when extending a feature whose audit log was started under the legacy canon.

Profile Mappings (Single Source of Truth)

Behavior Flow

Mode Detection

• No argument -> Mode 1 (Interactive Selection)
• Argument is a preset name (lean, standard, thorough, exhaustive, inherit) -> Mode 2 (Quick Switch)
• Argument is custom -> Mode 3 (Custom Builder)

Mode 1: Interactive Selection (no argument)

Step 1: Welcome

Read .nwave/des-config.json. If missing or .nwave/ directory absent -> error: "No nWave config directory found. Run nwave install first."

If JSON is invalid -> backup as .nwave/des-config.json.bak, reset config to {}, note: "Config was corrupted. Backed up and reset."

Display current profile (from config.rigor.profile) or "none set" if absent.

Brief explanation: "Rigor profiles control how much quality infrastructure nWave applies per wave: agent models, review depth, TDD phases, mutation testing. Higher rigor = better guarantees, higher token cost."

Step 1.5: Scope Selection

Display the current project rigor (from .nwave/des-config.json) and current global rigor (from ~/.nwave/global-config.json, if it exists).

Ask via AskUserQuestion:

Where do you want to save this configuration?

Options:

1. Globally (~/.nwave/global-config.json) — applies to all projects without their own rigor
2. This project only (.nwave/des-config.json) — overrides global for this project

Store the user's choice as {scope} and the corresponding file path as {target_file}:

• If global: {target_file} = ~/.nwave/global-config.json
• If project: {target_file} = .nwave/des-config.json

Step 2: Comparison Table

Display this table:

+-----------+--------+----------+----------+------------+---------+
|           | lean   | standard | thorough | exhaustive | inherit |
+-----------+--------+----------+----------+------------+---------+
| Agent     | haiku  | sonnet   | opus     | opus       | *yours* |
| Reviewer  | --     | haiku    | sonnet   | opus       | haiku   |
| Review    | no     | yes      | double   | double     | yes     |
| TDD       | R->G   | 5-phase  | 5-phase  | 5-phase    | 5-phase |
| Refactor  | no     | yes      | yes      | yes        | yes     |
| Mutation  | no     | no       | no       | yes        | no      |
+-----------+--------+----------+----------+------------+---------+
| Est. cost | lowest | moderate | higher   | highest    | varies  |
| Est. time | fast   | moderate | slower   | slowest    | varies  |
+-----------+--------+----------+----------+------------+---------+

Mark "standard" as [recommended]. Below the table, note: "Or choose custom to configure each setting individually. Type inherit to use your current session model."

Step 3: User Selection

Ask user to select via AskUserQuestion (4 options + Other for inherit/custom):

1. standard [recommended]
2. lean
3. thorough
4. exhaustive

Note in the question text: "Type 'custom' to build your own profile, or 'inherit' to use your session model."

If user selects or types "custom" -> jump to Mode 3 (Custom Builder). If user types "inherit" -> proceed with inherit profile to Step 4.

Step 4: Detail View

Show the detail view for the selected profile. Render in a code block for visual clarity.

lean:

WHAT YOU GET:
  - Haiku agent (fastest, cheapest)
  - RED -> GREEN TDD (skip PREPARE, RED_ACCEPTANCE, COMMIT phases)

WHAT YOU LOSE:
  - No code review
  - No PREPARE phase (no test fixture setup)
  - No RED_ACCEPTANCE phase (no acceptance tests)
  - No COMMIT phase (no refactoring pass)
  - No mutation testing

WHEN TO USE:
  Config changes, documentation, simple bug fixes, spikes/prototypes

ESTIMATED IMPACT:
  Lowest token cost | Fastest per step

standard [recommended]:

WHAT YOU GET:
  - Sonnet agent (balanced quality/speed)
  - Full 5-phase TDD (PREPARE -> RED_ACCEPTANCE -> RED_UNIT -> GREEN -> COMMIT)
  - Haiku reviewer (cost-effective review)
  - Refactoring pass in COMMIT phase

WHAT'S NOT INCLUDED:
  - No double review (single pass only)
  - No mutation testing
  - Not opus-level reasoning

WHEN TO USE:
  Most development work — features, integrations, refactoring

ESTIMATED IMPACT:
  Moderate token cost | Moderate time per step

thorough:

WHAT YOU GET:
  - Opus agent (strongest reasoning)
  - Sonnet reviewer (deeper review analysis)
  - Double review (two independent review passes)
  - Full 5-phase TDD
  - Refactoring pass

WHAT IT COSTS:
  Higher token cost | Slower per step

WHEN TO USE:
  Critical features, security-sensitive code, public APIs, complex algorithms

exhaustive:

WHAT YOU GET:
  - Opus agent and opus reviewer (strongest at every stage)
  - Double review (two independent review passes)
  - Full 5-phase TDD
  - Refactoring pass
  - Mutation testing (>= 80% kill rate gate)

WHAT IT COSTS:
  Highest token cost | Slowest per step

WHEN TO USE:
  Critical production systems, compliance-sensitive code, long-lived core modules

inherit:

WHAT YOU GET:
  - Your session model for agents (nWave inherits, does not override)
  - Haiku reviewer
  - Full 5-phase TDD
  - Single review pass
  - Refactoring pass

WHAT THIS MEANS:
  nWave respects your model choice and controls the process around it.
  If your session runs opus, agents get opus. If sonnet, agents get sonnet.

WHEN TO USE:
  When you have strong opinions about which model to use,
  or your organization controls model selection externally.

Step 5: Confirm

Ask user to confirm via AskUserQuestion:

1. Yes, apply this profile
2. No, go back to selection (return to Step 2)
3. Cancel (exit without saving)

Step 6: Save to Config

1. If {scope} is global AND the directory ~/.nwave/ does not exist, create it with parents=True
2. Read {target_file} (handle missing file or corrupt JSON: start with {})
3. Parse JSON
4. Set config["rigor"] to the full profile object:

   {
     "profile": "{selected}",
     "agent_model": "...",
     "reviewer_model": "...",
     "tdd_phases": [...],
     "review_enabled": true/false,
     "double_review": true/false,
     "mutation_enabled": true/false,
     "refactor_pass": true/false
   }

5. Write back to {target_file}, preserving all other top-level keys (audit_logging_enabled, skill_tracking, update_check, etc.)

Step 7: Summary

Display all resolved settings:

Rigor profile saved: {name}

  Resolved settings:
  +-----------------------+---------------------------------------------------+
  | agent_model           | {value}                                           |
  | reviewer_model        | {value}                                           |
  | tdd_phases            | {value}                                           |
  | review_enabled        | {value}                                           |
  | double_review         | {value}                                           |
  | mutation_enabled      | {value}                                           |
  | refactor_pass         | {value}                                           |
  +-----------------------+---------------------------------------------------+

  Config: {target_file} ({scope})
  All wave commands will use these settings.

Mode 2: Quick Switch (with argument)

Step 1: Validate Argument

If argument is not one of: lean, standard, thorough, exhaustive, custom, inherit -> error: "Unknown profile '{name}'. Available: lean, standard, thorough, exhaustive, custom, inherit"

If argument is custom -> redirect to Mode 3 (Custom Builder).

Read .nwave/des-config.json. If missing -> same error as Mode 1 Step 1.

Step 1.5: Scope Selection

Same as Mode 1 Step 1.5. Ask scope question, store {scope} and {target_file}.

Step 2: Show Diff

Display what changes from current profile to target profile:

Switching from {current} -> {target}:

  agent_model:      sonnet -> haiku
  reviewer_model:   haiku -> skip
  review_enabled:   true -> false
  tdd_phases:       5-phase -> R->G
  refactor_pass:    true -> false
  mutation_enabled: (unchanged) false

If downgrading (moving to a less rigorous profile), highlight what user will lose:

You will LOSE:
  - Code review (reviewer_model: skip)
  - PREPARE, RED_ACCEPTANCE, COMMIT phases
  - Refactoring pass

If no current profile is set, show the target profile settings without diff.

Step 3: Confirm

Ask user to confirm via AskUserQuestion:

1. Yes, switch to {target}
2. No, keep current profile

Step 4: Save + Summary

Same as Mode 1 Steps 6 and 7. Uses {target_file} from Step 1.5.

Mode 3: Custom Builder (/nw-rigor custom or selected from interactive)

Build a profile setting by setting. Each question uses AskUserQuestion with sensible defaults (standard values pre-selected). After all questions, show summary and confirm.