Nw NewSAFE
AI agents that guide you from idea to working code, with you in control at every step.
Overview
AI agents that guide you from idea to working code, with you in control at every step.
1a864ec81cdeOBSERVED · 2026-10-14What it tells the agent
The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.
---
name: nw-new
description: "Guided wizard to start a new feature. Asks what you want to build, recommends the right starting wave, and launches it."
user-invocable: false
argument-hint: '[feature-description] - Example: "Add rate limiting to the API gateway"'
---
# NW-NEW: Start a New Feature
**Wave**: CROSS_WAVE (entry point)
**Agent**: Main Instance (self — wizard)
**Command**: `/nw-new`
## Overview
Conversational wizard that asks the user to describe their feature|classifies it|recommends a starting wave|launches it. Eliminates need to understand the 6-wave pipeline before using nWave.
You (the main Claude instance) run this wizard directly. No subagent delegation.
## Behavior Flow
### Step 1: Feature Description Intake
Ask the user to describe what they want to build. If provided as argument, use that.
If vague (fewer than 3 meaningful words or unclassifiable):
- Ask: **What system** is being changed?
- Ask: **What problem** are you solving?
- Ask: **Who benefits** from this change?
- Provide example: "Add rate limiting to prevent API abuse"
Do NOT proceed until you have a clear, actionable description.
### Step 2: Feature ID Derivation
Derive feature ID per rules in `~/.claude/nWave/skills/common/wizard-shared-rules.md` (section: Feature ID Derivation).
Examples: "Add rate limiting to the API gateway" -> `rate-limiting-api-gateway` | "OAuth2 upgrade" -> `oauth2-upgrade` | "Implement a real-time notification system with WebSocket support for mobile and desktop clients" -> `real-time-notification-system-websocket`
Show derived ID via AskUserQuestion. Allow override with custom value.
### Step 3: Name Conflict Check
Check if `docs/feature/{feature-id}/` exists. If so, offer via AskUserQuestion:
1. **Continue that project** — switch to `/nw-continue`
2. **Start fresh with different name** — ask for distinguishing name
3. **Archive and restart** — move to `docs/feature/{feature-id}-archived-{date}/`
### Step 4: Clarifying Questions
Use AskUserQuestion:
**Q1: New or existing behavior?**
- New functionality that doesn't exist yet
- Changing/improving existing functionality
- Fixing a bug or regression
**Q2: Requirements readiness?**
- Clear requirements in my head but nothing written
- Rough idea, need to explore further
- Haven't validated whether problem is worth solving
### Step 5: Feature Classification
Based on description and answers, classify:
- **User-facing** — UI/UX visible to end users
- **Backend** — APIs, services, data processing
- **Infrastructure** — DevOps, CI/CD, tooling
- **Cross-cutting** — Spans multiple layers (auth, logging, etc.)
Show classification for user confirmation.
### Step 6: Greenfield vs Brownfield Detection
Scan filesystem:
- Check if `src/` or equivalent has code
- Check if `docs/feature/` has prior feature directories
- Check for existing test directories
No source code and no prior features -> **greenfield** | Otherwise -> **brownfield**
### Step 7: Wave Recommendation
Decision tree:
```
IF "fixing a bug":
-> /nw-root-why ("Investigate the root cause first")
IF "haven't validated the problem":
-> /nw-discover ("Validate the problem space before building")
IF "rough idea, need to explore":
-> /nw-discuss ("Define requirements and acceptance criteria")
IF "clear requirements, nothing written":
-> /nw-discuss ("Formalize requirements into user stories")
IF existing DISCUSS artifacts found:
-> /nw-design ("Requirements exist, design the architecture")
IF existing DESIGN artifacts found:
-> /nw-distill ("Architecture exists, create acceptance tests")
IF all prior waves complete:
-> /nw-deliver ("Ready for implementation")
DEFAULT:
-> /nw-discuss ("Start by defining what to build")
```
Show recommendation with rationale via AskUserQuestion: recommended wave command|why this wave (one sentence)|what it produces.
### Step 8: Launch
After user confirms, create project directory:
```bash
mkdir -p docs/feature/{feature-id}
```
Invoke recommended wave command by reading its task file and following instructions, passing feature ID as argument.
## Error Handling
| Error | Response |
|-------|----------|
| Vague description (< 3 meaningful words) | Ask follow-up questions with example |
| Name conflict with existing project | Offer continue/rename/archive options |
| User cannot classify feature type | Default to "cross-cutting", note uncertainty |
| No clear wave recommendation | Default to DISCUSS with explanation |
## Success Criteria
- [ ] User described feature in plain language
- [ ] Project ID derived and confirmed by user
- [ ] No name conflicts (or resolved)
- [ ] Feature classified by type
- [ ] Starting wave recommended with rationale
- [ ] User confirmed recommendation
- [ ] Wave command launched with correct feature ID
## Examples
### Example 1: Greenfield backend feature
```
/nw-new "Add rate limiting to the API gateway"
```
Derives `rate-limiting-api-gateway`, detects no prior artifacts (greenfield), asks clarifying questions. User says "new functionality, clear requirements." Recommends DISCUSS, launches `/nw-discuss "rate-limiting-api-gateway"`.
### Example 2: Bug fix
```
/nw-new "Fix authentication timeout errors"
```
Detects "fix" in description. User confirms bug. Recommends `/nw-root-why "authentication timeout errors"`.
### Example 3: Unclear problem
```
/nw-new "Build a customer feedback system"
```
User says they haven't validated whether customers want this. Recommends DISCOVER, launches `/nw-discover "customer-feedback-system"`.Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | NA |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- none-observed
- Network
- none-observed
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (0)
No findings outside the package's declared scope.
Gates applied: no_behavioural_pass.
1a864ec81cdefull audit observations/trust-audit/skill/nwave-ai__nw-new.json · Report an issue / request a re-scanAudit history
Every audit this skill has had.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-14 | 1a864ec81cde | SAFE | B | 89 | first audit |
Questions
What does the Nw New skill do?
AI agents that guide you from idea to working code, with you in control at every step.
Is Nw New safe to install?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean skill reads B.
What can Nw New access on my machine?
The audit observed no filesystem, network or shell use at all in its source.
How current is this page?
The grade is for one exact copy of the source (1a864ec81cde), read on 2026-10-14. The repository is watched, and a new audit runs when it changes — this is the first audit.