help
Skill15 starsChanged 4 months agoMITView on GitHub
Show KERNEL help. The unified skill system, state operations, agents, current plugin status. Triggers: help, how, what, skills, commands.
76 other projects use a version of this.
Adopted by
Most recently taken up by
- iskron-ai/skillstheir own version, first seen Nov 8, 2026
- nearmap/shipyardtheir own version, first seen Nov 8, 2026
- tristanphvn/keeltheir own version, first seen Nov 8, 2026
- Jasenpan1987/multi-codetheir own version, first seen Nov 8, 2026
- codewhale-hq/codewhale-plugin-marketplacetheir own version, first seen Nov 8, 2026
- onetapstudiogames/1f3d9-citylifetheir own version, first seen Nov 8, 2026
Used with
Projects with it also use create-stories, story-done and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
lilangMaxSkillcreate-stories
Break a single epic into implementable story files. Reads the epic, its GDD, governing ADRs, and control manifest. Each story embeds its GDD requirement TR-ID, ADR guidance, acceptance criteria, story type, and test evidence path. Run after /create-epics for each epic.
Also in 9 projects that use this
- lilangMaxSkill
story-done
End-of-story completion review. Reads the story file, verifies each acceptance criterion against the implementation, checks for GDD/ADR deviations, prompts code review, updates story status to Complete, and surfaces the next ready story from the sprint.
Also in 9 projects that use this
- +10 more from lilangMax
Other versions76 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 76
- Donchitos/Claude-Code-Game-Studios★ 27ksame name
- parcadei/Continuous-Claude★ 3.9ksame name
- parcadei/Continuous-Claude-v2★ 3.9ksame name
- parcadei/Continuous-Claude-v3★ 3.9ksame name
- agenticnotetaking/arscontexta★ 3.5ksame name
- indranilbanerjee/digital-marketing-pro★ 934same name
- waynesutton/markdown-site★ 628same name
- bitwize-music-studio/claude-ai-music-skills★ 589same name
- andrew-shwetzer/career-ops-plugin★ 501same name
- andrew-shwetzer/career-ops-plugin-do-not-fork-currently-updating-v2-★ 500same name
66 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository30 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- llms.txtllms.txt
- app-devSkill
- architectureSkill
- buildSkill
- checkpointSkill
- context-mgmtSkill
- debugSkill
- dreamSkill
- evalSkill
- experimentSkill
- forgeSkill
- frontendSkill
- governance-syncSkill
- handoffSkill
- human-passSkill
- ingestSkill
- initSkill
- knowledge-graphSkill
- landing-pageSkill
- marketing-siteSkill
- metricsSkill
- orchestrationSkill
- qualitySkill
- retrospectiveSkill
- reviewSkill
- shipSkill
- simplifySkill
- tearitapartSkill
The file
We split this skill into the rules it gives an agent
2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 57 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 28
Before showing help, check the actual state of the plugin as loaded in your context:
- Distinctive: line 161
Run retrospective: After several sessions, synthesize what you've learned </tips>
“Common” is measured: kept by at least 6 in 10 of the 57 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
<skill id="help"><purpose>
Quick reference for KERNEL v9.11.0.
One primitive: skills (methodology, workflows, state transitions, validators,
operators). Agents, manifests, philosophy, and current plugin status.
</purpose><host_invocation>
Claude Code invokes skills as /kernel:<name>. Codex invokes them as
$kernel:<name>. Tables below use Claude Code syntax; Codex users replace the
leading slash with a dollar sign.
</host_invocation>
<on_start> Before showing help, check the actual state of the plugin as loaded in your context:
- Is the session-start hook output visible? (Look for "# KERNEL" at conversation start)
- Which profile is active? (local, github, github-oss, github-production)
- Are there active contracts or pending reviews in AgentDB?
- Report what you see, don't just recite docs.
agentdb recall "<feature> <subsystem> <files/symbols> <error/outcome>" --global</on_start>
<getting_started>
| What you want | What to type |
| Set up a new project | /kernel:init |
| Start working on anything | /kernel:ingest + describe task |
| Run autonomously overnight | /kernel:forge + describe goal |
| Explore approaches creatively | /kernel:dream + describe problem |
| Debug something broken | /kernel:debug + describe symptom |
| Save progress before stopping | /kernel:handoff |
| </getting_started> |
<skills_reference>
Core Workflow (kind: workflow)
| Skill | Purpose | When to use |
/kernel:ingest | Guided entry, classify, scope, execute | Default for any task. Human confirms each phase. |
/kernel:forge | Autonomous engine, heat/hammer/quench/anneal | Run overnight. Iterates until antifragile or reports why not. |
/kernel:dream | Creative exploration, 3 perspectives + stress test | When you need competing approaches before committing. |
/kernel:debug | Diagnosis before prescription | Bugs, regressions, or mapping a refactor. Bug mode and refactor mode. |
/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. |
Quality & Review (kind: validator)
| Skill | Purpose | When to use |
/kernel:validate | Pre-commit gates, build, lint, test, security | Before every commit. Blocks on failure. |
/kernel:tearitapart | Critical pre-implementation review | Before tier 2+ work. Verdict: PROCEED/REVISE/RETHINK. |
/kernel:review | Code review for PRs or staged changes | Before merging. >80% confidence threshold. |
Learning & Observability
| Skill | Purpose | When to use |
/kernel:retrospective | Cross-session learning synthesis | When 5+ learnings accumulated. Clusters, deduplicates, promotes patterns into project hooks/agents/skills. |
/kernel:metrics | Observability dashboard | Check session stats, agent success rates, hook health, learning utilization. |
State Operations (kind: state_transition) — json-first
| Skill | Emits | When to use |
/kernel:handoff | kernel.handoff/v1 json manifest | Ending a session. Pins provenance, decisions, phases, context policy + budget. |
/kernel:checkpoint | kernel.checkpoint/v1 json manifest | MID-task save: safe context reset without full handoff ceremony. |
/kernel:retrospective | kernel.retrospective-result/v1 mutation record | 5+ learnings accumulated; promotes via artifact ladder. |
/kernel:ingest (resume) | kernel.context-receipt/v1 | Resuming: validates the manifest, checks divergence, compiles bounded context. |
Manifest CLI: orchestration/manifest/kernel-manifest (validate | latest | divergence |
preflight | compile | resume | activate | deactivate). Context policies: sealed (hook-blocked
forbidden globs) | bounded (extra loads ledgered) | advisory. Details: docs/MIGRATION-8.md.
Setup
| Skill | Purpose | When to use |
/kernel:init | Confirm Vaults, create missing data dirs and three safe helper links | Once per machine, or helper-link recovery. |
/kernel:help | This help | When you need a reminder. |
| </skills_reference> |
<flow_examples> Typical flows, skills chain together:
New feature: ingest → (dream if complex) → tearitapart → execute → validate → review → handoff
Bug fix: diagnose → ingest (with diagnosis) → validate → review
Overnight autonomous: forge (runs heat/hammer/quench/anneal loop, ships when antifragile)
End of session: retrospective (if learnings accumulated) → handoff
Long task, context filling up: checkpoint → /clear → ingest (auto-discovers the checkpoint, resumes bounded) </flow_examples>
<tiers>
Tier by reversibility x silence x blast radius; file count is only a weak hint.| Tier | Risk profile | Your Role | Skills involved |
| 1 | easy to undo, loud if wrong | Execute directly | ingest → validate |
| 2 | persistent or moderately quiet | Orchestrate, spawn surgeon | ingest → tearitapart → validate → review |
| 3 | hard to undo, quiet, wide blast | Orchestrate, surgeon + adversary | ingest → tearitapart → validate → review |
</tiers><agents>
These are Claude Code agent definitions. Codex does not register the files as native
agents; KERNEL applies the role contracts to available Codex subagents.| Agent | Role |
| Surgeon | Minimal diff implementation. Only touches contract-listed files. |
| Adversary | QA, assumes broken, finds edge cases, proves with evidence. |
| Reviewer | Code review with APPROVE/REQUEST CHANGES/COMMENT verdict. |
| Researcher | Finds proven solutions and anti-patterns before coding. |
| Scout | Codebase reconnaissance, maps structure, detects tooling. |
| Validator | Pre-commit quality gate, build, types, lint, tests, security. |
| Dreamer | Multi-perspective debate, minimalist/maximalist/pragmatist. |
</agents><lifecycle>
Claude Code runs the full declared plugin lifecycle. Codex runs supported synchronous
events, including SessionStart and write guards, but skips asynchronous command hooks
and has no plugin SessionEnd event. In Codex, use `$kernel:handoff` explicitly when
durable end-of-session state is required.
</lifecycle><philosophy>
<principle id="research_first">Research anti-patterns before solutions. Most problems are already solved.</principle>
<principle id="tests_first">Define success before coding. Tests before implementation.</principle>
<principle id="agentdb">Recall task context before acting. Recall again as the task changes. Write at end.</principle>
<principle id="big5">Big 5: input validation, edge cases, error handling, duplication, complexity.</principle>
<principle id="builtin">Built-in beats library. Library beats custom. Prove you need complexity.</principle>
</philosophy><tips>
- **Be specific**: "Add rate limiting to /api/upload" > "make it more secure"
- **Use the right skill**: diagnose for bugs, dream for design, ingest for everything else
- **Check metrics**: `/kernel:metrics` shows if learnings are being used or ignored
- **Save deliberately**: `/kernel:checkpoint` mid-task or `/kernel:handoff` before a session boundary
- **Run retrospective**: After several sessions, synthesize what you've learned
</tips></skill>Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.