1---2name: help3description: Show KERNEL help. The unified skill system, state operations, agents, current plugin status. Triggers: help, how, what, skills, commands.4---56<skill id="help">78<purpose>9Quick reference for KERNEL v9.10.3.10One primitive: skills (methodology, workflows, state transitions, validators,11operators). Agents, manifests, philosophy, and current plugin status.12</purpose>1314<host_invocation>15Claude Code invokes skills as `/kernel:<name>`. Codex invokes them as16`$kernel:<name>`. Tables below use Claude Code syntax; Codex users replace the17leading slash with a dollar sign.18</host_invocation>1920<on_start>21Before showing help, check the actual state of the plugin as loaded in your context:221. Is the session-start hook output visible? (Look for "# KERNEL" at conversation start)232. Which profile is active? (local, github, github-oss, github-production)243. Are there active contracts or pending reviews in AgentDB?254. Report what you see, don't just recite docs.2627```bash28agentdb recall "<feature> <subsystem> <files/symbols> <error/outcome>" --global29```30</on_start>3132<getting_started>33| What you want | What to type |34|---------------|--------------|35| Set up a new project | `/kernel:init` |36| Start working on anything | `/kernel:ingest` + describe task |37| Run autonomously overnight | `/kernel:forge` + describe goal |38| Explore approaches creatively | `/kernel:dream` + describe problem |39| Debug something broken | `/kernel:diagnose` + describe symptom |40| Save progress before stopping | `/kernel:handoff` |41</getting_started>4243<skills_reference>44## Core Workflow (kind: workflow)4546| Skill | Purpose | When to use |47|-------|---------|-------------|48| `/kernel:ingest` | Guided entry, classify, scope, execute | Default for any task. Human confirms each phase. |49| `/kernel:forge` | Autonomous engine, heat/hammer/quench/anneal | Run overnight. Iterates until antifragile or reports why not. |50| `/kernel:dream` | Creative exploration, 3 perspectives + stress test | When you need competing approaches before committing. |51| `/kernel:diagnose` | Systematic debugging + refactor analysis | Bugs, regressions, or before refactoring. Diagnosis before prescription. |52| `/kernel:simplify` | Cut cyclomatic complexity with AST-aware budgets, regression diffs, and a project gate | Jungle code, god functions, after any AI-written branchy function. A number, not an opinion. |5354## Quality & Review (kind: validator)5556| Skill | Purpose | When to use |57|-------|---------|-------------|58| `/kernel:validate` | Pre-commit gates, build, lint, test, security | Before every commit. Blocks on failure. |59| `/kernel:tearitapart` | Critical pre-implementation review | Before tier 2+ work. Verdict: PROCEED/REVISE/RETHINK. |60| `/kernel:review` | Code review for PRs or staged changes | Before merging. >80% confidence threshold. |6162## Learning & Observability6364| Skill | Purpose | When to use |65|-------|---------|-------------|66| `/kernel:retrospective` | Cross-session learning synthesis | When 5+ learnings accumulated. Clusters, deduplicates, promotes patterns into project hooks/agents/skills. |67| `/kernel:metrics` | Observability dashboard | Check session stats, agent success rates, hook health, learning utilization. |6869## State Operations (kind: state_transition) — json-first7071| Skill | Emits | When to use |72|-------|-------|-------------|73| `/kernel:handoff` | `kernel.handoff/v1` json manifest | Ending a session. Pins provenance, decisions, phases, context policy + budget. |74| `/kernel:checkpoint` | `kernel.checkpoint/v1` json manifest | MID-task save: safe context reset without full handoff ceremony. |75| `/kernel:retrospective` | `kernel.retrospective-result/v1` mutation record | 5+ learnings accumulated; promotes via artifact ladder. |76| `/kernel:ingest` (resume) | `kernel.context-receipt/v1` | Resuming: validates the manifest, checks divergence, compiles bounded context. |7778Manifest CLI: `orchestration/manifest/kernel-manifest` (validate | latest | divergence |79preflight | compile | resume | activate | deactivate). Context policies: sealed (hook-blocked80forbidden globs) | bounded (extra loads ledgered) | advisory. Details: docs/MIGRATION-8.md.8182## Setup8384| Skill | Purpose | When to use |85|-------|---------|-------------|86| `/kernel:init` | Confirm Vaults, create missing data dirs and three safe helper links | Once per machine, or helper-link recovery. |87| `/kernel:help` | This help | When you need a reminder. |88</skills_reference>8990<flow_examples>91Typical flows, skills chain together:9293**New feature:**94 ingest → (dream if complex) → tearitapart → execute → validate → review → handoff9596**Bug fix:**97 diagnose → ingest (with diagnosis) → validate → review9899**Overnight autonomous:**100 forge (runs heat/hammer/quench/anneal loop, ships when antifragile)101102**End of session:**103 retrospective (if learnings accumulated) → handoff104105**Long task, context filling up:**106 checkpoint → /clear → ingest (auto-discovers the checkpoint, resumes bounded)107</flow_examples>108109<tiers>110Tier by reversibility x silence x blast radius; file count is only a weak hint.111112| Tier | Risk profile | Your Role | Skills involved |113|------|--------------|-----------|-------------------|114| 1 | easy to undo, loud if wrong | Execute directly | ingest → validate |115| 2 | persistent or moderately quiet | Orchestrate, spawn surgeon | ingest → tearitapart → validate → review |116| 3 | hard to undo, quiet, wide blast | Orchestrate, surgeon + adversary | ingest → tearitapart → validate → review |117</tiers>118119<agents>120These are Claude Code agent definitions. Codex does not register the files as native121agents; KERNEL applies the role contracts to available Codex subagents.122123| Agent | Role |124|-------|------|125| **Surgeon** | Minimal diff implementation. Only touches contract-listed files. |126| **Adversary** | QA, assumes broken, finds edge cases, proves with evidence. |127| **Reviewer** | Code review with APPROVE/REQUEST CHANGES/COMMENT verdict. |128| **Researcher** | Finds proven solutions and anti-patterns before coding. |129| **Scout** | Codebase reconnaissance, maps structure, detects tooling. |130| **Validator** | Pre-commit quality gate, build, types, lint, tests, security. |131| **Dreamer** | Multi-perspective debate, minimalist/maximalist/pragmatist. |132</agents>133134<lifecycle>135Claude Code runs the full declared plugin lifecycle. Codex runs supported synchronous136events, including SessionStart and write guards, but skips asynchronous command hooks137and has no plugin SessionEnd event. In Codex, use `$kernel:handoff` explicitly when138durable end-of-session state is required.139</lifecycle>140141<philosophy>142<principle id="research_first">Research anti-patterns before solutions. Most problems are already solved.</principle>143<principle id="tests_first">Define success before coding. Tests before implementation.</principle>144<principle id="agentdb">Recall task context before acting. Recall again as the task changes. Write at end.</principle>145<principle id="big5">Big 5: input validation, edge cases, error handling, duplication, complexity.</principle>146<principle id="builtin">Built-in beats library. Library beats custom. Prove you need complexity.</principle>147</philosophy>148149<tips>150- **Be specific**: "Add rate limiting to /api/upload" > "make it more secure"151- **Use the right skill**: diagnose for bugs, dream for design, ingest for everything else152- **Check metrics**: `/kernel:metrics` shows if learnings are being used or ignored153- **Save deliberately**: `/kernel:checkpoint` mid-task or `/kernel:handoff` before a session boundary154- **Run retrospective**: After several sessions, synthesize what you've learned155</tips>156157</skill>
hashgraph-online