Autonomous Agents

Axiom includes 38 autonomous agents that automatically detect and diagnose common iOS development issues.

What Are Agents?

Agents are autonomous problem-solvers that:

• Trigger automatically based on keywords in your conversation
• Run independently with their own model and tools
• Scan your codebase for specific anti-patterns
• Provide actionable fixes with file:line references and code examples

How to Use Agents

Natural language (recommended) — Just describe what you want:

• "Check my code for accessibility issues" → accessibility-auditor triggers
• "Scan for memory leaks" → memory-auditor triggers
• "My SwiftUI app has janky scrolling" → swiftui-performance-analyzer triggers
• "Review for Swift 6 concurrency violations" → concurrency-auditor triggers
• "Check Core Data safety" → core-data-auditor triggers
• "Find Liquid Glass adoption opportunities" → liquid-glass-auditor triggers
• "Scan for deprecated networking APIs" → networking-auditor triggers
• "Review my in-app purchase implementation" → iap-auditor triggers
• "Implement in-app purchases" → iap-implementation triggers
• "My build is failing" → build-fixer triggers
• "My builds are slow" → build-optimizer triggers
• "Run a health check on my project" → health-check triggers
• "Scan everything for issues" → health-check triggers
• "Check my navigation architecture" → swiftui-nav-auditor triggers
• "Validate my App Store screenshots" → screenshot-validator triggers
• "Take a screenshot to verify this fix" → simulator-tester triggers

Explicit commands — For direct invocation:

/axiom:audit-accessibility
/axiom:audit-memory
/axiom:audit-swiftui-performance
/axiom:audit-concurrency
/axiom:audit-core-data
/axiom:audit-iap
/axiom:audit-liquid-glass
/axiom:audit-networking
/axiom:audit-swiftui-nav
/axiom:audit-icloud
/axiom:audit-storage
/axiom:fix-build
/axiom:optimize-build
/axiom:audit screenshots
/axiom:test-simulator
/axiom:health-check

Agent Categories

Project-Wide

• health-check — Orchestrates multiple specialized auditors in parallel, deduplicates findings, and produces a unified project health report with executive summary

Build & Environment

• build-fixer — Automatically diagnoses and fixes Xcode build failures using environment-first diagnostics (zombie processes, Derived Data, simulator state, SPM cache)
• build-optimizer — Scans for build performance optimizations (compilation mode, architecture settings, build phase scripts, type checking bottlenecks) with measurable time savings
• spm-conflict-resolver — Analyzes Package.swift and Package.resolved to diagnose and resolve Swift Package Manager dependency conflicts

Code Quality

• accessibility-auditor — Scans for VoiceOver label issues, Dynamic Type violations, color contrast failures, touch target sizes, WCAG compliance problems
• codable-auditor — Detects Codable anti-patterns (manual JSON building, try? swallowing errors, JSONSerialization usage) and date handling issues
• concurrency-auditor — Detects Swift 6 strict concurrency violations (missing @MainActor, unsafe Task captures, Sendable violations, actor isolation problems)
• energy-auditor — Scans for energy anti-patterns (timer abuse, polling, continuous location, animation leaks, background mode misuse)
• memory-auditor — Finds 6 common memory leak patterns (timers, observers, closures, delegates, view callbacks, PhotoKit accumulation)
• swift-performance-analyzer — Detects Swift performance anti-patterns (ARC overhead, unspecialized generics, collection inefficiencies, actor isolation costs)
• textkit-auditor — Scans for TextKit 1 fallback triggers, deprecated glyph APIs, missing Writing Tools support

UI & Design

• liquid-glass-auditor — Identifies iOS 26+ Liquid Glass adoption opportunities (glass effects, toolbar improvements, search patterns, migration from old blur effects)
• swiftui-architecture-auditor — Scans SwiftUI architecture (logic in view bodies, async boundary violations, property wrapper misuse, testability gaps)
• swiftui-layout-auditor — Scans SwiftUI layout code for GeometryReader misuse, deprecated screen APIs, hardcoded breakpoints, identity loss, missing lazy containers
• swiftui-performance-analyzer — Detects SwiftUI performance anti-patterns (expensive operations in view bodies, missing lazy loading, unnecessary updates, navigation performance issues)
• swiftui-nav-auditor — Scans SwiftUI navigation architecture (missing NavigationPath, deep link gaps, state restoration issues, wrong container usage, type safety problems)
• ux-flow-auditor — Detects UX dead ends, dismiss traps, buried CTAs, missing empty/loading/error states, and accessibility dead ends

Persistence & Storage

• core-data-auditor — Scans for schema migration risks, thread-confinement violations, N+1 query patterns, production data loss risks, performance issues
• database-schema-auditor — Scans database migration and schema code for unsafe ALTER TABLE patterns, DROP operations, missing idempotency, foreign key misuse
• icloud-auditor — Scans for iCloud integration issues (missing NSFileCoordinator, unsafe CloudKit error handling, missing entitlement checks, SwiftData + CloudKit anti-patterns)
• storage-auditor — Detects file storage mistakes (files in wrong locations, missing backup exclusions, missing file protection, storage anti-patterns causing data loss and backup bloat)
• swiftdata-auditor — Scans SwiftData code for struct models, missing VersionedSchema, relationship defaults, background context misuse, N+1 patterns

Integration

• camera-auditor — Scans for camera, video, and audio capture issues including deprecated APIs, missing interruption handlers, threading violations
• foundation-models-auditor — Scans Foundation Models code for missing availability checks, main thread blocking, manual JSON parsing, session lifecycle issues
• networking-auditor — Scans for deprecated networking APIs (SCNetworkReachability, CFSocket, NSStream) and anti-patterns (reachability checks, hardcoded IPs, missing error handling)
• iap-auditor — Audits existing IAP code for missing transaction.finish() calls, weak receipt validation, missing restore functionality, subscription status tracking issues, and StoreKit testing configuration gaps
• iap-implementation — Implements complete StoreKit 2 IAP solution with testing-first workflow (.storekit configuration, centralized StoreManager, transaction handling, subscription management, restore purchases)

Shipping

• screenshot-validator — AI-powered visual inspection of App Store screenshots for dimension validation, placeholder text detection, debug artifact scanning, competitor references, and content completeness
• security-privacy-scanner — Scans for API keys in code, insecure @AppStorage usage, missing Privacy Manifests, ATS violations, and logging sensitive data

Testing

• performance-profiler — Automated performance profiling via xctrace CLI (CPU Profiler, Allocations, Leaks, SwiftUI, Swift Tasks)
• simulator-tester — Automated simulator testing with visual verification (screenshots, video, location simulation, push notifications, permissions, deep links, log analysis) for closed-loop debugging
• test-debugger — Closed-loop test debugging: analyzes failures, suggests fixes, re-runs tests until passing
• test-failure-analyzer — Diagnoses flaky tests, race conditions, and tests that pass locally but fail in CI
• test-runner — Runs XCUITests, parses .xcresult bundles, provides structured results with failure analysis
• testing-auditor — Scans for flaky test patterns, shared mutable state, missing assertions, Swift Testing migration opportunities

Games

• spritekit-auditor — Scans SpriteKit code for physics bitmask issues, draw call waste, node accumulation, action memory leaks, coordinate confusion, and touch handling bugs

Misc

• crash-analyzer — Parses crash reports (.ips, .crash), checks symbolication, categorizes by crash pattern, generates actionable diagnostics
• modernization-helper — Scans for legacy patterns and provides migration paths to iOS 17/18 (ObservableObject to @Observable, etc.)

Why Agents?

Before (Commands):

• User must remember /axiom:audit-accessibility
• Manual invocation every time
• Duplication between command and skill implementations

After (Agents):

• Natural language: "check accessibility"
• Automatic triggering based on context
• One source of truth, zero duplication
• Scales better (agents = files + commands, not duplicated implementations)

Agent Architecture

All agents:

• Use haiku model for fast execution (<1 second scans)
• Provide file:line references for easy fixing
• Include severity ratings (CRITICAL/HIGH/MEDIUM/LOW)
• Show before/after code examples
• Recommend testing strategies to verify fixes

Browse All Agents

Select an agent from the sidebar to see its full documentation, detection patterns, and fix recommendations.