For your AI editor

Your AI editor already knows shilp-sutra.

One install. Four files. Every release.

We ship an installable skill, two flavours of llms.txt, and AGENTS.md with every release so your coding agent doesn't guess at the library. Install once, never paste docs into chat again.

One installRun this in your terminal.

bash

curl -fsSL https://raw.githubusercontent.com/devalok-design/shilp-sutra/main/skills/shilp-sutra/install.sh | bash

Works with

Every tool that speaks Agent Skills.

Claude Code
Anthropic agent. Skill loads natively.
Cursor
Skill works via Agent Skills standard.
Codex (Codex IDE)
Skill works via Agent Skills standard.
Aider
Reads llms.txt + AGENTS.md.
Continue.dev
Reads llms.txt directly.
Any other
agentskills.io-compatible tool.

What ships in the box

Four files. Pick the one your tool reads.

Agent Skill

Anthropic-format installable skill. Ships every setup playbook, component reference, theming guide, and troubleshoot tree.

curl -fsSL https://raw.githubusercontent.com/devalok-design/shilp-sutra/main/skills/shilp-sutra/install.sh | bash

llms.txt

Concise cheatsheet, 660 lines. Setup playbook, peer-dep matrix, breaking changes, design rules. Inject this into any chat.

curl https://raw.githubusercontent.com/devalok-design/shilp-sutra/main/packages/core/llms.txt

llms-full.txt

Exhaustive per-component reference, 6,900 lines. Every prop, every variant, every example. Reach for this when llms.txt is too brief.

curl https://raw.githubusercontent.com/devalok-design/shilp-sutra/main/packages/core/llms-full.txt

AGENTS.md

Agent contract at the repo root. The rules every coding agent should follow when working on shilp-sutra or a consumer of it.

cat AGENTS.md

Why this exists

The library outpaces training data.

Coding agents trained pre-2026 hallucinate component APIs that never existed. Skill + llms.txt give your agent the current shape of the library, including what changed in the last release.

Spec

Built on the open standard.

The skill follows the agentskills.io specification. Any tool that adopts the standard reads our skill the same way.