documentation benchmark

docs.cdp.coinbase.com benchmark.

Report for

docs.cdp.coinbase.comdocs.cdp.coinbase.com

Agents should be able to find the right page, fetch the content, and follow the instructions without much extra guesswork. This documentation benchmark shows where agents can use these docs cleanly and where the biggest points of friction still are.

Last checked: May 10, 2026Category: DocsAlot runsBenchmark type: documentation benchmarkSource: DocsAlot run

Benchmark score

100Grade Excellent

This report passes 16 of 17 checks, with 0 warnings, 1 failed checks, and 0 skipped checks still shaping the next score lift.

Checks passed: 16/17
Warnings: 0
Failed checks: 1

Key takeaways

What stands out in this benchmark.

Start with what already works, then move to the fixes most likely to lift the score.

Working now

What already works.

LLMS TXT exists
Markdown parity exists
LLMS TXT is structured

Improve next

What to improve next.

List key constraints and requirements

Check results

Full score breakdown

Open each category to review the checks behind the score and see exactly where agents still hit friction.

Discovery and delivery

3 passed

Can an agent find the docs and fetch readable text?

This area is already easy for agents to work through.

LLMS TXT exists

Found llms.txt at https://docs.cdp.coinbase.com/llms.txt (7871 chars)

LLMS Full exists

Found llms-full.txt at https://docs.cdp.coinbase.com/llms-full.txt (12325315 chars)

Markdown parity exists

5/5 sampled pages have valid .md versions. Pages sourced from llms.txt. Checked: ✓ https://docs.cdp.coinbase.com/index.md, ✓ https://docs.cdp.coinbase.com/get-started/quickstart.md, ✓ https://docs.cdp.coinbase.com/get-started/authentication/overview.md, ✓ https://docs.cdp.coinbase.com/get-started/authentication/jwt-authentication.md, ✓ https://docs.cdp.coinbase.com/get-started/authentication/security-best-practices.md

Content structure

4 passed

Can an agent understand how the page is organized?

This area is already easy for agents to work through.

LLMS TXT is structured

✓ Has H1 title; ✓ Has blockquote description; ✓ Has 9 H2 sections; ✓ Has list items with links

Links include descriptions

51/51 links have descriptions

Docs sections are structured

Found 9 recognized section types in H2 headings. Your sections: "getting started", "cdp cli & mcp", "api reference", "guides", "ai & agent tooling", "core api concepts", "sdks", "tutorials", "optional"

Optional sections are separated

Has an ## Optional section - AI tools can skip this content when context window is limited

Task readiness

4 passed1 failed

Can an agent follow the instructions without guessing?

This area is usable, but it still adds avoidable friction.

Product description is clear

Blockquote description has 15 words

Constraints are documented

No constraints, requirements, or important notes found in opening

Integrations are documented

Found integration references: integration, extension, connect

Error handling is documented

Found error/troubleshooting references: error, troubleshoot

Examples are included

Found example/tutorial section

Access and recovery

5 passed

Can an agent complete the task without getting blocked?

This area is already easy for agents to work through.

Markdown routes resolve

5/5 markdown pages return valid markdown content

Pages respond quickly

Average response time: 250ms across 7 requests

Content is readable without JS

Content is present in the initial HTML response (server-rendered)

Markdown negotiation works

3/3 pages return markdown when requested with Accept: text/markdown. Checked: ✓ https://docs.cdp.coinbase.com/index, ✓ https://docs.cdp.coinbase.com/get-started/quickstart, ✓ https://docs.cdp.coinbase.com/get-started/authentication/overview

Markdown output stays clean

Markdown content is clean (no navigation, footer, or cookie elements)

Next steps