Tencent RTC LLM Docs

This directory provides LLM-oriented documentation entry files.

How to use

If you want an LLM to read Tencent RTC docs, provide the raw text URL instead of a repository web page URL.

Raw text URLs are usually better for LLM retrieval because they:

• contain less navigation and page chrome noise
• are easier to parse as plain text
• produce cleaner context with lower token waste

Recommended entry files

Chinese

• Index: zh/llms.txt
• Full product docs: zh/llms-full-*.txt

Recommended raw URL pattern:

• https://gitee.com/Tencent-RTC/docs/raw/main/zh/llms.txt
• https://gitee.com/Tencent-RTC/docs/raw/main/zh/llms-full-<product>.txt

English

• Index: en/llms.txt
• Full product docs: en/llms-full-*.txt

Recommended raw URL pattern:

• https://raw.githubusercontent.com/Tencent-RTC/docs/main/en/llms.txt
• https://raw.githubusercontent.com/Tencent-RTC/docs/main/en/llms-full-<product>.txt

Which file should I give to the LLM?

• Use llms.txt when you want the LLM to discover the right document first.
• Use llms-full-*.txt when you already know the product and want deeper retrieval.

Quick choice

• If you do not know which document to use, give the LLM llms.txt.
• If you already know the product, give the LLM llms-full-<product>.txt.
• If you already know the exact page, give the LLM the raw URL of that Markdown file.

Single-page Markdown

If you only want the LLM to read one document, provide the raw URL of that Markdown file directly.

Examples:

• Chinese page: https://gitee.com/Tencent-RTC/docs/raw/main/zh/webrtc/plugins/watermark.md
• English page: https://raw.githubusercontent.com/Tencent-RTC/docs/main/en/webrtc/plugins/watermark.md

Use a single-page Markdown URL when the question is already focused on one feature, one API, or one guide.

Do not use the rendered repository page URL when a raw Markdown URL is available.

Notes

• Prefer raw text URLs over rendered repository pages.
• zh/** and en/** contain the source Markdown files.
• llms.txt is the lightweight index.
• llms-full-*.txt contains aggregated product-level retrieval docs.