Kernel Claude

Eval-Driven Development (EDD) for AI workflows. pass@k metrics, capability evals, regression evals. Triggers: eval, edd, pass@k, capability, regression, benchmark.

Show KERNEL help. The unified skill system, state operations, agents, current plugin status. Triggers: help, how, what, skills, commands.

Deep creative exploration engine. Competing perspectives, stress-tested by a 4-persona council, scored by integrity. The approach that survives attack wins — not the one that sounds best.

Autonomous development with experimental self-correction. Iterates through quality gates until antifragile or blocked.

Entry point for new or resumed work. Researches and scopes new tasks; validates and resumes handoff/checkpoint manifests. Triggers: start, begin, do, implement, build, fix, create, resume, continue.

Mobile/web app build pipeline: fastlane-first local builds, store submission, pre-submission checklists. Triggers: app, mobile, store submission, build, deploy app, fastlane, expo, react native, flutter.

Save provenance, decisions, phases, context policy, and next steps as kernel.handoff/v1. Triggers: handoff, save, pause, context, continue later.

Observability dashboard. Session stats, agent tracking, hook performance, learning health. Triggers: metrics, stats, dashboard, observability.

AI code quality checks. The Big 5: input validation, edge cases, error handling, duplication, complexity. Triggers: quality, big 5, ai code, review, validate.

Systematic debugging and refactor analysis. Diagnosis before prescription. Bug mode: reproduce, trace, isolate, hypothesize, diagnose. Refactor mode: map, trace deps, measure coupling, risks, diagnose.

Context-led frontend design. Derives art direction from product, audience, brand, content, and the existing system; avoids generic defaults. Triggers: frontend, UI, CSS, styling, visual, theme, component, layout, responsive, accessibility, design system, aesthetic, animation, motion, abyss, spatial, verdant, substrate, ember, arctic, void, patina, signal, kinetic-type, haunted-machine, cursor-field, real-weight, one-surface, liquid-material, held-breath, grand-tour.

Simplify code, plans, and systems without removing required outcomes or features; for code, lower cyclomatic complexity with AST-aware measurement and regression gates. Triggers: simplify, refactor, complexity, spaghetti, overengineered.

Save bounded progress as kernel.checkpoint/v1 for exact resume after reset. Triggers: checkpoint, save progress, context reset, compact soon, long task.

Experiment engine for development rules. Treats rules as hypotheses, designs falsifiable tests, updates confidence, and graduates or kills rules. Triggers: experiment, hypothesis, prove, test rule, validate methodology, scientific, evidence.

Guide real-build acceptance: literal taps, real inputs, expected outcomes, worst case first, one sitting. Triggers: testflight, human pass, acceptance test, before we ship, hand it to the user, what should I test, release checklist, device test.

Critical pre-implementation review. Find what AI breaks. Verdict: PROCEED, REVISE, or RETHINK. Triggers: review plan, tear apart, critique, analyze.

System architecture and design patterns. Modular design, interface stability, dependency management, AI-code health nexus. Triggers: architecture, design, structure, modules, dependencies, coupling, system design.

Context engineering and token management. Compaction strategies, progressive disclosure, structured note-taking via AgentDB. Triggers: tokens, compaction, memory, handoff, summarize, context window.

Explicit landing-page build and deploy operator. Loads marketing and frontend judgment, scaffolds the smallest suitable site, verifies it rendered, deploys to the configured target.

Multi-agent orchestration. Lane contracts, worker-model doctrine, fault tolerance, worktree isolation. Triggers: orchestrate, coordinate, agents, parallel, spawn, contract, tier 2, tier 3.

Audit whether prior learnings actually fired, then convert this run's recurrences into installed mechanisms. Refuses to emit a finding without an install path and a firing test. Triggers: retrospective, reflect, what did we learn, patterns, synthesis, post-mortem, why does this keep happening.

Honest marketing-site method: audience, offer, proof, objections, privacy, art direction, handoff. Triggers: marketing site, landing page, company website, product website, portfolio, campaign page, client website, conversion, positioning, offer, value proposition, CTA, social proof, testimonial, privacy policy.

Audit or explicitly synchronize Claude and Codex repository governance without overwriting conflicts. Triggers: governance sync, CLAUDE.md, AGENTS.md, instruction mirror.

Build + keep-fresh + query a deterministic code knowledge graph to cut agent orientation-token cost. Triggers: knowledge graph, graphify, code graph, god nodes, orientation cost, token bill, map the codebase, what connects X to Y, callers of, blast radius.