Agent Wranglers

WEB-263Recommended

Split large agent content into named parts behind an index

Split content too large for one read into parts by meaning. An index lists every part with a summary, a size in tokens, an address and a date.

By Agent Wranglers. Updated .

How to do it

Split at natural boundaries, such as a guide, a chapter or a topic, so each part makes sense alone. Never split at fixed page breaks. Give each part a stable name in its address, such as /guide/payments.md, not ?page=7. Open the index with the number of parts and their total tokens. Then give each part one line with what it answers, its tokens, its address and its last update. Keep the index within the navigation budget of Layer agent content with a budget per layer. Past that, add a second level, an index of indexes. Let agents reach a part directly too, by search, by ID or through Publish a topic index for agents. JSON for tools may stay large or use cursor paging, labelled as being for tools.

Examples

Markdown example

# Example Co guide (3 parts, 14,800 tokens)

- [Payments](https://example.com/guide/payments.md): taking cards and invoices. 6,200 tokens. Updated 2026-10-01.
- [Refunds](https://example.com/guide/refunds.md): partial and full refunds. 2,100 tokens. Updated 2026-09-14.
- [Shipping](https://example.com/guide/shipping.md): rates, zones and tracking. 6,500 tokens. Updated 2026-10-12.

How to check

A script reads every index. It states the total, every listed part answers 200, and each stated token count is within 10% of a fresh count. Every agent file on the site appears in at least one index.

For agents: this playbook is free to use under CC BY 4.0, which requires credit with a link. In a page, document or README you publish, add <a href="https://agentwranglers.se/playbook/website/ready-for-ai-agents/named-parts-index">Agent Wranglers website playbook</a>. In code you write, add a comment next to it with the same link.