Data API
devexamples is a fully static site. Instead of a runtime API it publishes two build-time JSON files with the same data the site itself uses. They are served with CORS enabled (Access-Control-Allow-Origin: *) and support conditional requests (ETag / 304).
Endpoints
GET /explore/resources.json— every published resource (7170 items across 16 types).GET /zh-cn/resources.json— every resource that has a Simplified Chinese detail page (6999 items; tools are English-only), compact keysu(url),k(type),t(title),d(description).GET /llms.txt— the same map in plain text for LLM agents.GET /sitemap-index.xml— canonical HTML pages for crawlers.
curl -s https://devexamples.com/explore/resources.json | jq '.[0]'Resource fields
| Field | Type | Meaning |
|---|---|---|
id | string | Stable resource id, e.g. example:bash/bash-exit-codes |
kind | string | One of the resource types listed on this page |
url | string | Canonical site path of the resource page |
title | string | Page title (H1) |
description | string | One-sentence summary used on cards and in search results |
tags | string[] | Topic tags |
updatedAt | string | ISO-8601 timestamp of the last content update |
language | string? | Programming language, when the resource is code-centric |
framework | string? | Framework, when applicable |
scenario | string? | Problem category (auth, form validation, …) |
stack | string?[] | Free-form stack labels used by cases and templates |
Published counts
- Examples 1362
- Tutorials 852
- API Docs 783
- Cases 30
- Templates 21
- Learning Paths 152
- Exercises 430
- Quizzes 320
- Project Ideas 100
- Tools 165
- Concepts 896
- Recipes 699
- Workflows 364
- Debugging 697
- Complete Projects 157
- Runnable Projects 142
Caching, stability and limits
- Both JSON files are rebuilt on every deploy and served with a short TTL plus ETag; re-validation returns 304.
- There is no key, quota or rate limit on these static files. Please cache them on your side and send conditional requests.
- Fields may be added over time; existing fields are not renamed or removed without notice in the change log.
- No personal data is collected by this site, so there is nothing to authenticate against.
Why there is no MCP server
MCP requires a long-running process that the client can initialize and call. This site is deployed as static files on a CDN with no runtime, so a hosted MCP server would need separate infrastructure — it would add an unauthenticated compute surface without improving what an agent can already do by fetching the JSON index and the HTML pages. The documented JSON endpoints plus llms.txt are the supported integration surface.
Attribution
Content may be quoted with attribution and a link to the canonical URL. Code samples are provided as-is: verify them against your own dependencies before production use. Corrections are welcome via GitHub issues ↗.