audit

Files (2)

• LIMITATIONS.md
• SKILL.md

SKILL.md (first lines)

# Audit Skill

Analyze feature completeness from the configured perspectives, identify gaps, and record them in a bounded evidence artifact. Plan documents are written only on an authorized run. The loop is bounded by the audit's own named unresolved questions, not by "nothing is missed".

## Purpose

Force comprehensive feature discovery by analyzing from every angle until nothing is missed. Don't stop after surface-level analysis.

## Setup

Run the setup script to initialize the audit loop:

```bash
"${CLAUDE_PLUGIN_ROOT}/scripts/setup-audit.sh" $ARGUMENTS
```

This creates:
- `.fno/audit-loop.local.md` — Loop config with completion promise
- `.fno/audit-progress.txt` — Iteration-persistent progress tracker

If the setup script sets a completion promise, read it and follow its instructions.

## Options

| Option | Description |
|--------|-------------|
| `--max-iterations N` | Stop after N iterations (default: 20) |
| `--output-dir PATH` | Where to write plan documents when the run is authorized to produce plans (default: the plans dir the `fno do plan path` chain resolves) |
| `--perspectives LIST` | Comma-separated lens subset; the resolved list is the authority for "do not skip any" (default: all five - ux, pm, po, eng, integration) |

## Skills, loaded for their own act

- `/think` — load for structured design exploration a lens needs
- `/blueprint` — load for plan production on an authorized run
- `/tdd` — load for acceptance-criteria writing on planned features
- `/setup` — load for cross-project workspace context

## Process

### 1. Load Context (Every Iteration)

Each iteration is a FRESH context. Memory persists ONLY via:
- **Progress file** — `.fno/audit-progress.txt` (READ THIS FIRST!)
- **Created plans** — Plan folders you've already created
- **Git history** — Commits from previous iterations

### 2. Analyze Current State

Scan codebase and existing plans to build completeness matrix:

```markdown
## Current State Summary

| Area | Status | Completeness | Evidence |
|------|--------|--------------|----------|
| QR Sign-in Flow | ✅ Complete | 95% | src/routes/sign-in/, tests pass |
| Child Roster | ✅ Complete | 90% | CRUD works, missing bulk import |
| Ratio Monitoring | ⚠️ Partial | 80% | Real-time works, no forecasting |
| Notifications | ❌ Missing | 0% | No SMS delivery, no alerts |
```

**How to scan:**
1. Read existing plan documents in the output directory
2. Search codebase for implemented features
3. Run the project's verification command once to ground the matrix - not once per lens
4. Check for TODO comments and incomplete features

### 3. Multi-Perspective Gap Analysis

Analyze from every lens in the resolved `--perspectives` set (default: all five, including Integration Coherence). Do not skip a lens the run resolved.

#### UX Research Perspective

```markdown
## User Journey Gaps

| Journey | Gap | Impact | Priority |
|---------|-----|--------|----------|
| Parent First-Time | No onboarding flow | High friction | P1 |
| Staff Discovery | No setup guide | Confusion | P1 |
| Multi-Child Parent | One-at-a-time sign-in | Slow | P2 |

## Missing Edge Cases
- Custody changes: No same-day revocation
- Emergency contacts: Can't mark "no pickup"
- Late pickup: No alerts

## Accessibility Gaps
- Signature canvas needs keyboard alternative
- Color-only status indicators
```

#### Product Owner Perspective (INVEST Stories)

```markdown
## Must-Have (P1) - Blockers

**US-1: SMS Delivery**
As a parent, I need to receive OTP via SMS
so that I can verify my phone.

Acceptance Criteria:
- [ ] Twilio sends real SMS
- [ ] Handles delivery failures
- [ ] Rate limits (5/hour/phone)

## Should-Have (P2) - Adoption

**US-4: Multi-Child Sign-in**
As a parent with multiple children, I need to sign them together
so that drop-off is faster.

## Could-Have (P3) - Scale

**US-7: PWA**
As a parent, I want an app instead of QR each time...
```

#### Product Manager Perspective

From bllshttng/fno, Apache-2.0.

More from bllshttng/fno