RF SAFE · PUBLIC RECORD

Set up and update the archive

The same front end runs on this ChatGPT Site and in a regular website folder. On your PHP server, the importer can fetch the official comment record. This preview is a saved snapshot, not a continuously running collector.

1. Upload one folder

Extract hhs-rfi-wireless-radiation/ into your website. For example, it can appear at https://rfsafe.org/hhs-rfi-wireless-radiation/. No build command, Node.js process, database, or URL rewriting is needed. Serve it over HTTP or HTTPS; opening the HTML directly as a local file cannot load its JSON data.

2. Add your official API key

Get a free key at api.data.gov. The read API does not require permission to submit comments. Keep the key in the server’s REGULATIONS_API_KEY environment variable. Alternatively, copy server/config.example.php to server/config.local.php and enter it there. Never put the key in assets/app.js, a URL, a public JSON file, or an uploaded ZIP.

The importer requires PHP 8.0 or later with cURL and JSON, write access to the archive’s data folder, and outbound HTTPS access to Regulations.gov. The website itself only needs static-file hosting. Apache’s included server/.htaccess denies web access to the server folder. If you use Nginx, deny all web requests to this archive’s /server/ path in your host configuration before adding a key. Private history must never be publicly served.

3. Import new and changed comments

cd /path/to/public_html/hhs-rfi-wireless-radiation
php server/update.php sync --limit=250 --attachments

This is a command-line task. Run it through your hosting terminal, an authorized AI server connection, or a cron job. The web “Update this archive” button opens these instructions; it does not execute a server job or expose your key.

The updater discovers docket documents, paginates their public comments, and retrieves missing or changed records. It saves progress after each comment. Rerun the same command to continue after a quota limit or interrupted run. The limit controls how many comment details are retrieved in one run; listing pages and attachment downloads use additional requests. --attachments archives supported files up to 20 MB each when their official download is accessible. Unsupported, oversized, or unavailable files retain their official links.

Use --refresh periodically to recheck existing details and public profile changes. A complete listing that no longer includes a previously archived record withdraws its text from the public snapshot and moves cached source files into the protected server history. A partial listing never triggers that removal.

4. Ask AI to summarize the new records

Tell your AI: “Read the archive’s update skill, retrieve pending comments, and review new or changed submissions.” The updater’s initial topic suggestions use keywords; they are explicitly not AI reviews. Your AI reads the source, writes an attributed summary, selects a primary concern and additional topics, and supplies exact supporting passages.

php server/update.php apply /path/to/annotations.json
php server/update.php rebuild

Review batches must match the current source hash. If the source changes, its previous summary is withheld until reviewed again. Original text is never overwritten by a summary. Attachment-only submissions remain unreviewed until the attachment is read.

Optional: schedule the import

Your hosting control panel can run the importer daily or more frequently within your API quota. For example, after adjusting the path and configuring the environment:

15 */6 * * * /usr/bin/php /path/to/public_html/hhs-rfi-wireless-radiation/server/update.php sync --limit=250 --attachments

This example is not an installed schedule. Cron retrieves records and creates keyword suggestions; it does not perform AI review. The preview does not update merely because your server’s copy changed. Ask your AI to publish the refreshed snapshot when updating the ChatGPT Site.

What is in the folder?

Initial preview coverage

The initial official index contains 234 records. Six detail records were retrieved at evenly spaced positions in document-ID order to exercise the preview; five comment bodies have AI summaries. This is a development sample, not a representative sample of views. One retrieved submission consists of an attachment, and two others contain attachments; their three attachment files could not be downloaded in the build environment and have not been summarized. Their official file links are retained. The remaining 228 records have metadata only.

Bulk harvesting was not started with the demonstration API key. A personal key enables the PHP updater to retrieve the remaining content within your quota. A supplied API export can also be imported: php server/update.php import /path/to/export-folder. Include comments-list.json and individual detail JSON responses named with their official comment IDs.