SaneProcess Full Audit Skill

Trigger: /audit or "run audit" or "full audit"

Purpose

Comprehensive system audit using PARALLEL SUB-AGENTS for each section. This prevents context bloat and ensures deep, focused analysis.

CRITICAL: Each audit section MUST be run by a dedicated Task agent. Do NOT run checks inline - delegate to sub-agents.

Execution Flow

Step 1: Launch All Audit Agents IN PARALLEL

You MUST launch these Task agents in a SINGLE message with multiple tool calls. This runs them concurrently.

Launch these 6 agents IN PARALLEL (single message, multiple Task tool calls):

1. Task(subagent_type="Explore", prompt=<Q0_CONFIG_PROMPT>)
2. Task(subagent_type="Explore", prompt=<Q6_RELEASE_PROMPT>)
3. Task(subagent_type="Explore", prompt=<Q7_WEBSITE_PROMPT>)
4. Task(subagent_type="Explore", prompt=<Q8_SIGNING_PROMPT>)
5. Task(subagent_type="Explore", prompt=<Q9_SUPPORT_PROMPT>)
6. Task(subagent_type="Explore", prompt=<Q10_DOCS_PROMPT>)

Step 2: Run Ruby Validation (Background)

While agents run, also execute the Ruby validation script:

ruby ~/SaneApps/infra/SaneProcess/scripts/validation_report.rb 2>&1

Step 3: Collect & Present Results

Wait for all agents to complete, then present unified report.

Sub-Agent Prompts (Intelligent Expert Personas)

Each sub-agent has:

1. •Expert Persona - Domain expertise to find gaps
2. •Phase 1: Baseline Checklist - Known items to verify
3. •Phase 2: Expert Gap Analysis - Find what checklist missed
4. •Phase 3: Output Report - Rating, gaps, suggestions

Q0: Config Consistency Agent

Persona: DevOps Config Guardian (15 yrs infrastructure experience)

Read: prompts/q0-config.md

Baseline Focus:

• •Deprecated plugins in settings.json
• •MCP server paths (npm vs local)
• •Hook configurations
• •Environment variable locations (.zprofile vs .zshrc)
• •Sister app consistency

Expert Gap Analysis:

• •Configuration drift and orphaned files
• •Security concerns in configs
• •Maintainability issues
• •Cross-platform readiness

Q6: Release Integrity Agent (CRITICAL)

Persona: Release Engineering Lead (12 yrs macOS distribution)

Read: prompts/q6-release.md

Baseline Focus:

• •Appcast.xml URLs return 200/302 (NOT 404)
• •GitHub releases exist for each app
• •DMG files downloadable
• •Sparkle signatures present
• •minimumSystemVersion not too restrictive

Expert Gap Analysis:

• •Release infrastructure resilience
• •Download reliability
• •Update experience for users
• •Release process gaps
• •NEVER suggests Homebrew

This catches the SaneBar disaster before customers complain.

Q7: Website Distribution Agent

Persona: Web Infrastructure Architect (10 yrs web reliability)

Read: prompts/q7-website.md

Baseline Focus:

• •All product domains accessible
• •SSL certificates valid
• •Download links work
• •No 5xx errors

Expert Gap Analysis:

• •SSL/TLS security
• •DNS configuration
• •CDN & performance
• •Website content accuracy
• •Legal compliance

Q8: Code Signing Agent

Persona: Apple Security Engineer (12 yrs macOS security)

Read: prompts/q8-signing.md

Baseline Focus:

• •Developer ID Application identity exists
• •Certificates not expired
• •Notarytool keychain profile works
• •App bundles properly signed
• •Notarization stapled

Expert Gap Analysis:

• •Certificate health and expiration
• •Hardened runtime and entitlements
• •DMG signing
• •Build process integrity
• •Archive management

Q9: Support Infrastructure Agent

Persona: Customer Success Architect (10 yrs support ops)

Read: prompts/q9-support.md

Baseline Focus:

• •Keychain has required API keys
• •Resend API responding
• •Sane-Mem service running
• •Support email working

Expert Gap Analysis:

• •Lemon Squeezy store per app
• •Email infrastructure
• •Support channels
• •Credential security
• •Monitoring & alerting
• •Customer data handling

Q10: Documentation Currency Agent

Persona: Technical Documentation Lead (12 yrs tech writing)

Read: prompts/q10-docs.md

Baseline Focus:

• •README mentions current version
• •CHANGELOG has latest release
• •SESSION_HANDOFF.md not stale
• •Version consistency across docs

Expert Gap Analysis:

• •User-facing documentation gaps
• •Developer documentation
• •Release documentation quality
• •Legal/policy documentation
• •Hidden/undocumented features

Report Format

After all agents complete, aggregate their reports into a unified view:

# Full System Audit Report

**Generated:** [timestamp]
**Verdict:** [WORKING / NEEDS ATTENTION / BROKEN PIPELINE]

---

## Completeness Scorecard

| Section | Expert | Score | Justification |
|---------|--------|-------|---------------|
| Q0 | DevOps Config Guardian | 7/10 | Hooks good, missing centralized config |
| Q6 | Release Engineering Lead | 8/10 | URLs work, no rollback plan |
| Q7 | Web Infrastructure Architect | 6/10 | Sites up, privacy policy missing |
| Q8 | Apple Security Engineer | 9/10 | Signing solid, archives not preserved |
| Q9 | Customer Success Architect | 5/10 | Keys present, Lemon Squeezy not set up |
| Q10 | Technical Documentation Lead | 6/10 | Versions match, hidden features undocumented |

**Overall Score: X/10** (average of all sections)

---

## Critical Issues (Customer-Facing)
| Section | Issue | Impact |
|---------|-------|--------|
| Q6 | Release URL 404 | Customers can't update |

## Expert-Identified Gaps (Beyond Checklist)
| Section | Gap | Risk Level | Recommendation |
|---------|-----|------------|----------------|
| Q9 | No Lemon Squeezy store for SaneHosts | High | Set up before launch |
| Q8 | Archives not preserved | Medium | Save .xcarchive files |

## Warnings
| Section | Issue | Recommendation |
|---------|-------|----------------|

## All Clear (Baseline)
- Q0: Config ✅
- Q7: Websites ✅

## Actions Required
1. [ ] [Critical] Fix release URL for SaneClip
2. [ ] [High] Set up Lemon Squeezy for SaneHosts
3. [ ] [Medium] Update README version

---

## Suggested Checklist Improvements

Based on expert analysis, add these checks to future audits:

| Section | Suggested Addition | Why |
|---------|-------------------|-----|
| Q8 | Archive preservation check | Can't recreate releases without them |
| Q9 | Per-app Lemon Squeezy verification | Caught missing payment infrastructure |

Key Rules

1. •CURRENT PROJECT ONLY - Audit the project in CWD, not all SaneApps
2. •PARALLEL EXECUTION - All 6 agents launch in ONE message
3. •NO INLINE CHECKS - Everything delegated to sub-agents
4. •FOCUSED CONTEXT - Each agent only sees its specific domain
5. •CUSTOMER-FIRST - Q6/Q7/Q8 issues are CRITICAL
6. •ACTIONABLE OUTPUT - Clear list of what to fix
7. •INTELLIGENT ANALYSIS - Agents find gaps BEYOND the checklist
8. •COMPLETENESS RATINGS - Every section rated 1-10 with justification
9. •CONTINUOUS IMPROVEMENT - Agents suggest checklist additions
10. •NO HOMEBREW - Never suggest Homebrew distribution. Ever.

Exception: User can explicitly request cross-project audit (e.g., "audit all apps").

Quick Mode

/audit --quick - Run only Q6 (Release) and Q8 (Signing) for fast pre-release check.

Why Sub-Agents?

This is how you designed it. This is how it runs.