Integrate in-app help in about an hour

Integrating HelpCCMS takes one request and one component.

GET https://www.helpccms.com/api/deploy/{collection}/{key}

The endpoint returns the published topic for a key as HTML and as plain text. No SDK or client package is required, and reading published content needs no token.

What a developer does once

1. Get the collection id. Open the collection in HelpCCMS and copy its id from the register. The collection id is an address, not a secret. Put it in your application configuration.

2. Add one fetch function.

const BASE = "https://www.helpccms.com/api/deploy";
const COLLECTION = process.env.HELP_COLLECTION;

export async function getHelp(key) {
  const response = await fetch(
    `${BASE}/${COLLECTION}/${encodeURIComponent(key)}`
  );

  if (!response.ok) return null;

  return response.json();
}

3. Render the response. A topic contains both html and text. Use html for a rich surface and text where only plain text fits.

export function Help({ helpKey }) {
  const help = use(getHelp(helpKey));

  if (!help) return null;

  return (
    <div dangerouslySetInnerHTML={{ __html: help.html }} />
  );
}

You decide where the content appears: a tooltip, a panel, inline help, an onboarding step or a support screen.

4. Handle missing content. A key without published content returns 404. Treat missing help as optional content, and do not let it break the surrounding interface.

That is the integration.

Replacing hardcoded help

A coding agent can extract the explanatory text already in your codebase: tooltips, error explanations, onboarding text, empty-state help and help panels.

For each item, record a stable help key, the current text, and where it is used in the code. Import that file into HelpCCMS.

The imported text is not rewritten. During the migration, users keep seeing the same wording. Then replace each hardcoded string with a request for its help key.

The import format has the JSON shape and an extraction prompt you can hand to the agent.

After the integration

Developers no longer change code for ordinary help-text updates. A writer edits and publishes the topic in HelpCCMS, your application requests the current published version, and it arrives within the cache window. No application release is required.

Where the boundary is

HelpCCMS manages the content. Your application controls when it appears, where it appears, who sees it, and how it is rendered. HelpCCMS does not inject a user interface into your application.

That keeps presentation and product logic in your code while the content stays separately managed. The security page covers what public delivery means for access.

Before you start

Caching. Responses include an ETag and can be cached for up to five minutes. If your framework already caches requests, you can use that cache.

Client or server. The delivery API is public and CORS is open, so you can request content from the browser or from your server. The collection id is not a credential.

One key, one topic. A help key identifies one published topic within a collection. Duplicate keys are rejected before delivery, so your application never has to resolve an ambiguity.

Prove the integration

1. Publish a topic with the key test.hello.
2. Request it from your application.
3. Render the returned content.
4. Change the topic in HelpCCMS and publish again.
5. Reload the application after the cache window.

If the new text appears without an application deploy, the integration is complete.

The API reference has the response fields, the error codes, the caching headers and the versioning policy. Pro publications can also include machine-readable semantic metadata.

← HelpCCMS