Tomjesch | AI Blog Writer

Aprašymas

Tomjesch AI Blog Writer adds an „Add with AI” button to selected post types in the WordPress admin list screens. From there, you can describe the post you want, and the plugin will call your configured AI provider to generate a draft (title + HTML content) and open it directly in the editor for review.

The plugin is provider-agnostic and currently supports:

  • OpenAI Chat Completions API (GPT-4.x, GPT-4o, GPT-5 families, etc.)
  • Anthropic Claude Messages API (Claude 4.x Sonnet and Haiku variants)

All AI calls are made directly from your WordPress site to the provider’s HTTPS API using your own API key. No data is sent to any server operated by the plugin author.

Key features

  • „Add with AI” action on supported post types in the admin list table
  • Provider-specific model selection for OpenAI and Anthropic
  • Base (master) prompt for tone, structure, SEO and style
  • Connection test from the settings page, including basic token limit hints
  • Robust response normalization (handles JSON, embedded JSON, labeled text, HTML, and plain text)
  • Optional debug view showing the most recent generations (prompts + token usage)

SaaS and external services

This plugin is an interface to third-party AI services. To function, it requires you to configure an API key for at least one provider:

  • OpenAI – https://platform.openai.com/
  • Anthropic Claude – https://console.anthropic.com/

These services may be paid and are governed by their own terms of use and privacy policies. This plugin does not alter or override those terms; it simply sends your prompts and instructions to the provider you choose and displays the generated content.

Privacy

This plugin:

  • Sends the base prompt and the per-request user prompt to your configured AI provider over HTTPS when you test the connection or generate a post.
  • Does not send any data to servers controlled by the plugin author.
  • Stores recent prompts and token usage locally in your WordPress database for debugging purposes only.

Debug log

When you generate a post, the following data is stored in a small rotating debug log:

  • Timestamp
  • Post type
  • Provider and model name
  • User prompt (truncated to 4,000 characters)
  • Base prompt (truncated to 4,000 characters)
  • Token usage information where the provider reports it
  • A small excerpt of the raw provider response
  • A normalized title excerpt and normalization metadata (detected format + any repairs applied)

Only the 20 most recent entries are kept. You can review them on the „AI Blog Writer Debug” page under Settings, and you can clear the entire history with the „Clear debug history” button on that page.

If you consider prompts or base prompts to contain personal data, you should clear the debug history regularly and avoid including personal data in prompts.

API keys

API keys can be stored in one of two ways:

  • In the plugin settings (stored in the WordPress options table).
  • In wp-config.php using the constant TOMJESCH_AI_BLOG_WRITER_API_KEY, which is recommended for extra security.

When the plugin is uninstalled from the Plugins screen, it automatically deletes its options, including any API key stored in the database and all debug entries.

If you choose to define an API key in wp-config.php, that key is stored in code, not in the database. When you stop using the plugin or rotate keys, you should also remove or update the TOMJESCH_AI_BLOG_WRITER_API_KEY constant in wp-config.php.

Diegimas

  1. Upload the tomjesch-ai-blog-writer folder to the /wp-content/plugins/ directory, or install the plugin through the WordPress Plugins screen.
  2. Activate the plugin through the „Plugins” screen in WordPress.
  3. Go to Settings AI Blog Writer:
    • Choose your provider (OpenAI or Anthropic).
    • Enter your API key (or define it in wp-config.php).
    • Select the default model for that provider.
    • Configure a base prompt if desired.
    • Select which post types should have the „Add with AI” button.
  4. (Optional) Use the Connection Test section on the settings page to verify that the API key and model work.

Once configured, visit the list view for a supported post type (for example, Posts). Next to the standard „Add New” button, you will see an „Add with AI” button. Click it to open the prompt modal and generate a draft.

DUK

Does this plugin send data to any server operated by the author?

No. All outbound HTTP requests are sent directly from your WordPress site to the AI provider you configure (OpenAI or Anthropic), using their official HTTPS APIs. The plugin does not proxy requests through any server controlled by the author.

What data is sent to the AI provider?

For each generation request, the plugin sends:

  • The base (master) prompt from the settings page (if configured).
  • The free-text description you type into the „Generate new post with AI” modal.
  • Minimal context about the post type (for example, its label) to help the model understand what you are creating.

The AI provider returns generated JSON containing a title and HTML body, which the plugin uses to create a draft.

Is there a way to clear prompts from the debug history?

Yes. Go to Settings AI Blog Writer Debug and click the „Clear debug history” button. This removes all stored debug entries from the database.

How are API keys stored, and how can I remove them?

  • If you enter an API key on the settings page, it is stored in the WordPress options table and is automatically removed when you uninstall the plugin.
  • If you define TOMJESCH_AI_BLOG_WRITER_API_KEY in wp-config.php, the key is stored in that file instead of the database. To remove it, edit wp-config.php and delete or change the constant definition.

Atsiliepimai

Įskiepis neturi atsiliepimų.

Programuotojai ir komandos nariai

“Tomjesch | AI Blog Writer” yra atviro kodo programa. Prie jos sukūrimo prisidėję žmonės surašyti toliau.

Autoriai

Pakeitimų istorija

  • 1.0.2 =
  • Improve robustness of generated output parsing/normalization (supports multiple response formats and common wrapper/escaping issues).
  • Expand debug entries with normalization metadata while keeping a small rotating history.

  • 1.0.1 =

  • Refactor admin inline scripts to enqueued JS assets (performance/compatibility); keep JSON-LD unchanged.

1.0.0

  • Initial release.