# Install the Sema Skill > One-click install of Sema language support for your coding agent (Claude Code, Cortex/OpenClaw, Codex, or any LLM). No model is trained on Sema yet — the skill teaches it. import { Tabs, TabItem, Card, CardGrid, LinkButton } from '@astrojs/starlight/components'; No large language model has been trained on Sema — it's a new language. So out of the box, an assistant will guess at the syntax and get the neurosymbolic parts wrong. The **Sema language skill** fixes that: it's a compact, authoritative brief (the same one Cortex ships to its own agents) that teaches a model the toolchain, the core and neurosymbolic constructs, and the `edit → check → assure` workflow — so it can actually read and write correct Sema. Install it into your coding agent in one command. ## Quick install (auto-detect) Run this in your project. It detects your agent (Claude Code, Cortex/OpenClaw, Codex) and installs the skill in the right place: ```bash curl -fsSL https://sema.49.12.246.95.sslip.io/skills/install.sh | sh ``` Then **start a new agent session** so it picks up the skill. ## Per-agent instructions Install into the current project (`.claude/skills/`): ```bash curl -fsSL https://sema.49.12.246.95.sslip.io/skills/install.sh | sh -s -- claude ``` Or install globally for every project (`~/.claude/skills/`): ```bash curl -fsSL https://sema.49.12.246.95.sslip.io/skills/install.sh | sh -s -- global ``` Claude Code auto-discovers `SKILL.md` files under `.claude/skills/` and loads the Sema skill whenever you work on `.sema` files. Restart the session after installing. Cortex agents (and OpenClaw / OMP harnesses) read shared skills from `.agents/skills/`: ```bash curl -fsSL https://sema.49.12.246.95.sslip.io/skills/install.sh | sh -s -- cortex ``` If your harness generates a registry, run its sync afterward (e.g. `make agent-registry-sync`) so the skill is indexed for every CLI. Codex reads `AGENTS.md`. The installer drops the skill at `.codex/sema-language.md` and adds a pointer to `AGENTS.md`: ```bash curl -fsSL https://sema.49.12.246.95.sslip.io/skills/install.sh | sh -s -- codex ``` Every agent can use the raw Markdown. Download the skill and add it to your assistant's rules/instructions, or just paste it into the chat: ```bash curl -fsSL https://sema.49.12.246.95.sslip.io/skills/sema-language/SKILL.md -o SKILL.md ``` For the full language corpus, point the model at [`/llms.txt`](/llms.txt) (index) or [`/llms-full.txt`](/llms-full.txt) (everything), or use the **Open in ChatGPT / Claude** buttons at the top of any page. ## What the skill contains `sema check` / `run` / `assure` / `doc`, `SEMA_STRICT`, and how to verify Sema — plus `cargo` + grammar-drift for compiler work. Core (traits, enums, generics, effects) and neurosymbolic (`simulate`, `~=`, `semantic.*`, contracts, `policy`, `monitor`, `budget`, `protocol`). The mandatory `edit → check → assure` loop and the no-silent-no-ops ethos, so generated Sema is verified, not hoped for. The crates, the embedded stdlib, the examples corpus, and `LANGUAGE.md` — so the model knows where to look, not just what to type. Download SKILL.md :::note[Reproducible & inspectable] The installer only writes the skill file into your agent's skills directory (or `.codex/` + `AGENTS.md`). Read the script first if you like: [`install.sh`](/skills/install.sh). :::