llms.txt

llms.txt is a standard for exposing documentation to AI in a clean, plain-text form: a curated index of the docs (titles, links, descriptions) plus a full-text bundle. Think robots.txt, but for LLMs.

We generate these on every build, so they always match the current docs.

The main file (current v7 docs):

https://docs.thewidlarzgroup.com/react-native-video/llms.txt

Full-text and version-specific variants are in Files we publish below.

How it's actually used

Not by pasting it into a chat. In practice llms.txt is a Business-to-Agent format, and the real consumers are coding agents:

• Your coding agent fetches it on demand. Point Cursor, Claude Code, GitHub Copilot, Windsurf, Cline, or Aider at the URL (via Cursor's @Docs, or by listing it in a rules file like .cursor/rules/…, CLAUDE.md, or AGENTS.md), and the agent pulls llms.txt when it's relevant and follows the links it needs. It's URL-based: no copy-paste, no token bloat.
• Context7 (MCP) already has our docs. react-native-video is indexed on Context7, including the v7 documentation. With the Context7 MCP configured, just add use context7 to your prompt and your assistant gets up-to-date, version-specific react-native-video docs injected automatically, so you never touch a .txt file.
• The skill is the turnkey version. Install the skill and you get all of the above wrapped with automatic version detection. Nothing to wire up.
• Plain chat? Paste as a fallback. In a chat that can't fetch URLs (a basic ChatGPT/Claude window), paste llms-full.txt. Note: general web-crawling models don't reliably auto-load llms.txt, so treat this as a fallback, not the main path.

Honest take

IDE and coding agents use llms.txt today; the big web-search LLMs mostly don't yet, and it isn't an SEO trick. We publish it for the coding-agent workflow above.

Files we publish

Served from the docs root. Point your agent at llms.txt; the rest are variants for specific needs.

File	What it is
llms.txt	The one to link. Curated index of the current (v7) docs (titles, descriptions, links); the generated API reference is left out for signal.
llms-full.txt	The same curated docs as one full-text file for a single paste into a large-context chat.
llms-v7.txt / llms-v7-full.txt	v7 including the full API reference (bigger and noisier; use when you need exact TypeScript signatures).
llms-v6.txt / llms-v6-full.txt	Legacy v6 index / full text.

Set it up

• Cursor - add the URL to Cursor's indexed docs (Settings → Indexing & Docs, or @Docs → Add new doc in the chat; see Cursor's docs), or list it in a .cursor/rules/*.md file.
• Claude Code / Codex - add the URL to CLAUDE.md / AGENTS.md, or just install the skill.
• Copilot / Windsurf / Cline / Aider - reference the URL in the tool's rules or context file.
• Context7 - with the MCP set up, add use context7 to your prompt; it resolves react-native-video automatically.
• Plain chat (fallback) - paste llms-full.txt.

Which file?

• Link one URL: llms.txt (current v7, curated).
• Need exact API signatures: llms-v7.txt / llms-v7-full.txt.
• App is on v6: llms-v6.txt / llms-v6-full.txt. (Check package.json → react-native-video: 7.x incl. 7.0.0-beta.x is v7; 6.x is v6.)
• Want zero setup: install the skill, with version detection and routing built in.