Building Snaps with AI

This page collects everything an AI agent or automated tool needs to work with Farcaster Snaps.

Creating a snap with an agent

If you use a coding agent like Claude Code, you can ask it to install a skill that generates snaps from natural language:

install the farcaster-snap skill from https://docs.farcaster.xyz/snap/SKILL.md
make ... a Farcaster Snap poll asking users to pick their favorite variety of mole

The skill will:

1. Read the full snap spec
2. Generate valid snap code
3. Deploy it to a live URL

This is the fastest way to go from idea to working snap.

Machine-readable documentation

The full docs are available in machine-readable form:

• /llms.txt — all documentation concatenated into a single plain-text file, suitable for pasting into a context window or fetching at the start of a session
• any page of the docs can be requested with the Accept: text/markdown HTTP header to get Markdown-fromatted docs and save on tokens.
• /markdown-content/<slug> — individual pages as plain markdown (e.g. /markdown-content/spec-overview, /markdown-content/elements, etc)