clarc

Comprehensive accessibility audit — automated axe scan, keyboard navigation check, color contrast review, ARIA audit, and prioritized CRITICAL/HIGH/MEDIUM issue backlog

Add production observability to a service — structured logging, metrics endpoint, health checks, and Sentry error tracking. Detects project type and adds the right libraries and middleware.

Audit a clarc agent file for quality across 8 dimensions — instruction clarity, model appropriateness, tool coverage, trigger precision, exit criteria, examples, overlap detection, and safety guardrails. Produces a scored JSON report. Use /agent-audit --all to audit all agents at once.

Guided workshop for designing a multi-agent system — task decomposition, agent roles, orchestration pattern, pseudocode, and failure handling strategy

Review high-confidence instincts and promote them to agent-specific overlay files (~/.clarc/agent-instincts/). Closes the continuous-learning-v2 feedback loop by making agents smarter over time.

Reviews a single clarc agent file for quality across 8 dimensions — instruction clarity, model appropriateness, tool coverage, trigger precision, exit criteria, examples, overlap detection, and safety guardrails. Produces a scored JSON report. Use via /agent-audit or called by agent-system-reviewer during full system review.

Orchestrates a full clarc system review by synthesizing results from all component analyzers (agent-quality-reviewer, skill-depth-analyzer, command-auditor, hook-auditor), cross-component validators, and systemic effectiveness tools into a unified Priority Matrix with P0/P1/P2 classification. Uses Opus for deep architectural reasoning. Called by /system-review full.

Expert Android code reviewer specializing in Jetpack Compose best practices, Hilt DI scoping, Room migrations, ViewModel/UiState patterns, and Coroutine Dispatcher correctness. Use for all Kotlin/Android code changes. MUST BE USED for Android/Compose projects.

System design and architecture specialist. Use PROACTIVELY when users need HOW to structure the system — component boundaries, ADRs, C4/arc42 diagrams, scalability decisions. For task lists and implementation timelines (WHAT to build), use the planner agent instead.

Editorial critique for blog posts, essays, guides, and long-form articles — evaluates structure, opening strength, voice consistency, evidence quality, banned patterns, and SEO basics. Use after writing a first draft to get structured, actionable improvement feedback before publishing.

Advanced accessibility patterns — screen reader testing (NVDA/VoiceOver/TalkBack checklists), automated testing (axe-core+Playwright, jest-axe, @axe-core/react dev warnings), and accessible component implementations (modal, table, form, data visualization).

Web Accessibility (WCAG 2.2): key success criteria (contrast, keyboard, focus, ARIA), ARIA patterns (aria-label/labelledby/describedby, roles, live regions, state attributes), keyboard navigation (focus trap, skip links, roving tabindex), screen reader testing (NVDA/VoiceOver/TalkBack), automated testing (axe-core, axe-playwright, jest-axe), and accessible component patterns (modals, tables, forms, data visualizations).

Architecture Decision Records (ADRs): when to write one, the standard template, how to document rejected alternatives with real reasoning, how to supersede outdated ADRs, and how to maintain a living ADR index. The reference for technical documentation that actually gets read.

Decision framework for resolving conflicts between clarc agents — priority hierarchy, conflict classes, escalation protocol, and real-world examples

How the agent evolution system works — capturing instincts, promoting them to agent overlays, approval workflow, rollback, and the full continuous-learning flywheel pipeline

24 hooks across 7 events