Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
By ariaxhan
Orchestrate self-improving multi-agent coding workflows with persistent AgentDB memory, tiered task routing across 13 agents, and a scientific experiment engine that tests, validates, and evolves coding rules. Coordinate parallel git-isolated tasks without merge conflicts, remember failures across sessions, perform adversarial reviews, pre-commit validations, and autonomous TDD cycles.
npx claudepluginhub ariaxhan/kernel-claude --plugin kernelInitialize KERNEL globally. Sets up Vaults, symlinks, CLI, database.
Systematic debugging and refactor analysis. Diagnosis before prescription. Bug mode: reproduce, trace, isolate, hypothesize, diagnose. Refactor mode: map, trace deps, measure coupling, risks, diagnose.
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 experimentation engine. Every rule is a hypothesis. Seeds from CLAUDE.md, designs experiments, runs them, updates confidence, graduates or kills rules. One command. Walk away.
Autonomous development engine with experimental self-correction. Heats solutions, hammers through iteration, quenches with quality gates, experiments on its own output. Runs until antifragile or reports why it can't be.
QA - assume broken, find edge cases, prove with evidence
Cross-task intelligence. Detects dependencies, batches related work, spots systemic patterns.
Extracts patterns from human review decisions. Progressive rule promotion.
Whole-codebase reasoning with 1M context. Maps structure, dependencies, risk zones.
Structured post-mortem with telemetry evidence. Determines cause of death for failed work.
REST API design patterns. Resource naming, status codes, pagination, error responses, versioning. Triggers: api, rest, endpoint, route, http, status-code, pagination.
Mobile/web app build pipeline, EAS, store submission, pre-submission checklists. Triggers: app, mobile, EAS, store submission, build, deploy app, expo, react native, flutter.
System architecture and design patterns. Modular design, interface stability, dependency management, AI-code health nexus. Triggers: architecture, design, structure, modules, dependencies, coupling, system design.
Backend architecture patterns. Repository pattern, caching, queues, N+1 prevention, transactions. Triggers: backend, server, database, cache, redis, queue, repository, service.
Solution exploration and implementation. Generate 2-3 approaches, pick simplest. Never implement first idea. Triggers: build, implement, create, feature, add.
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
HelloAGENTS — The orchestration kernel that makes any AI CLI smarter. Adds intelligent routing, quality verification (Ralph Loop), safety guards, and notifications.
This skill should be used when the model's ROLE_TYPE is orchestrator and needs to delegate tasks to specialist sub-agents. Provides scientific delegation framework ensuring world-building context (WHERE, WHAT, WHY) while preserving agent autonomy in implementation decisions (HOW). Use when planning task delegation, structuring sub-agent prompts, or coordinating multi-agent workflows.
Agent orchestration harness for Claude Code — campaign persistence, fleet coordination, intent routing
OpenAgentsControl — multi-agent orchestration for Claude Code. Context-aware development with skills, subagents, parallel execution, and automated code review.
Multi-agent orchestration framework for Claude Code. Routes tasks to specialized Haiku/Sonnet subagents while Opus orchestrates — inspired by speculative decoding. Includes 10 specialized heads, environment preflight checks, and ~50% API cost reduction.
Core planning and workflow infrastructure for the Claudikins ecosystem
Uses power tools
Uses Bash, Write, or Edit tools
Uses power tools
Uses Bash, Write, or Edit tools
Share bugs, ideas, or general feedback.
Claude Code learns from itself.
Your agent forgets everything when you close it. KERNEL gives it persistent memory, multi-agent orchestration, and a scientific experiment engine that proves which rules actually work. Every session makes it smarter about YOUR project.
/plugin marketplace add ariaxhan/kernel-claude
/plugin install kernel
/kernel:init
ariaxhan/kernel-claude/init to set up memoryInstall via Claude Code or Claude Desktop first. Cursor shares the same plugin configuration automatically.
See docs/QUICKSTART.md for the full setup guide.
AgentDB remembers what worked, what broke, and where you left off — across every session. Not just logs. Weighted retrieval surfaces the learnings that matter (top 7% deliver 80% of value) while pruning what doesn't. Your agent stops repeating the same mistakes.
13 specialized agents route by complexity. Tier 1 (1-2 files) executes directly. Tier 2+ spawns surgeons to implement and an adversary to verify. The adversary checks coordination first (file overlap, scope drift, duplicate work) because our telemetry proved coordination failures are 4.3x more impactful than code bugs.
| Agent | Role |
|---|---|
| Surgeon | Minimal-diff implementation. Checkpoints to AgentDB. |
| Adversary | Coordination verification + code quality. Assumes broken until proven. |
| Reviewer | 11-phase code review, >80% confidence threshold. |
| Researcher | Finds solutions before building. Anti-patterns first. |
| Scout | Maps codebase structure, detects tooling, identifies risk. |
| Validator | Pre-commit: tests, lint, types, security scan. |
| Triage | Fast complexity classifier before expensive work. |
The experiment engine treats every rule as a hypothesis. It seeds them from your CLAUDE.md, designs experiments, runs them against AgentDB telemetry, and graduates rules that survive or kills rules that don't. 22 rules graduated from 107 hypotheses across 205 experiments. The forge command uses this: after building, it tempers — experiments on its own output, discovers emergent patterns, and self-corrects before shipping.
19 skills (testing, security, debug, api, backend, architecture, etc.) load when relevant — not at startup. Each is a methodology: HOW to approach a problem, not just tools to use.
Start with /ingest — the universal entry point. Reads memory, classifies your task, routes to the right agent.
/ingest add user authentication to the app
Run overnight with /forge — autonomous engine. Generates competing approaches, iterates against tests, adversarial review, experiments on output. Come back to shipped code.
Save with /handoff before closing. Next session, /ingest auto-resumes from where you left off.
Check with /validate before committing. Tests, lint, types, security.
Note: In Claude Code terminal, commands use the
kernel:prefix (/kernel:ingest). In Claude Desktop and Cursor, they appear without the prefix (/ingest).
| Terminal | Desktop/Cursor | What It Does |
|---|---|---|
/kernel:ingest | /ingest | Guided flow — classify, scope, execute. Auto-resumes from handoffs. |
/kernel:forge | /forge | Autonomous — heat/hammer/quench/temper/anneal until antifragile |
/kernel:experiment | /experiment | Run the hypothesis engine — seed, test, graduate, kill rules |
/kernel:dream | /dream | Creative exploration — 3 perspectives, 4-persona stress test |
/kernel:diagnose | /diagnose | Systematic debugging + refactor analysis before fixing |
/kernel:retrospective | /retrospective | Cross-session learning synthesis + pattern promotion |
/kernel:metrics | /metrics | Observability — sessions, agents, hooks, learnings |
/kernel:validate | /validate | Pre-commit quality gates |
/kernel:tearitapart | /tearitapart | Critical pre-implementation review |
/kernel:review | /review | Code review for PRs |
/kernel:handoff | /handoff | Save progress for next session |
/kernel:init | /init | Setup (run once per project) |
/kernel:help | /help | Show help |
/plugin marketplace update kernel-marketplace
/plugin update kernel@kernel-marketplace
/reload-plugins
/plugin and go to the Marketplaces tab