The kernel skill — zeq-os-mi-kernel
A skill is a markdown file with YAML frontmatter that an LLM client loads as a system prompt. The kernel skill is the boot sequence for any agent operating the Zeq engine: it pins the 1.287 Hz HulyaPulse, mandates KO42 on every call, and wires the seven-step protocol so every output carries the proper envelope and a ZeqProof. Where the MCP server gives an agent the instruments, the kernel skill gives it the rules.
One source of truth
The skill is served by every node from a single canonical source, so there's nothing to keep in sync — change it once and every install re-syncs on its next pull. Three endpoints, one truth:
| Endpoint | Format | When to use |
|---|---|---|
GET /api/kernel/prompt | text/plain | the raw boot sequence — paste into a system-prompt field |
GET /api/kernel/skill.md | text/markdown | the full SKILL.md with frontmatter — drop into a skills directory |
# grab the skill and hand it to any agent runtime
curl -s https://zeqsdk.com/api/kernel/skill.md
Why rules matter as much as tools
An agent with tools but no rules will guess: it'll invent operator IDs, skip the clock, and present a generated equation as a discovery. The kernel skill closes that gap — it tells the model the operators are a fixed vocabulary, that a generated equation wears a hypothesis stamp, and that every answer must carry a proof. Paired with the MCP tools, an agent both can call the framework and knows how it's meant to.
Read next
- MCP server — the instruments to pair with these rules.
- The AI harness — the consumer framing in full.
- The seven-step protocol — the contract the skill wires in.