=== MartinCV AI Blog Post ===
Contributors: martin7ba
Donate link: https://www.martincv.com/whp-donate
Tags: ai, openai, claude, gemini, post generator
Requires at least: 5.9
Tested up to: 7.1
Requires PHP: 7.4
Stable tag: 2.0.0
License: GPLv3 or later
License URI: http://www.gnu.org/licenses/gpl-3.0.html

Write complete blog posts — and matching images — from just a post title, using OpenAI, Anthropic Claude, Perplexity, or Google Gemini.

== Description ==

Type a post title, click **Generate Post**, and get a complete, well-structured blog post inserted straight into your editor — with AI-generated images saved to your Media Library. Version 2.0 turns this plugin from an OpenAI-only tool into a **full multi-provider AI writing assistant**:

- **OpenAI** — the GPT-5.6 family (Sol, Terra, Luna) plus still-supported classics like GPT-4.1, GPT-4o, and GPT-3.5 Turbo, with GPT Image 2 and DALL-E 3 for images.
- **Anthropic Claude** — Claude Opus 5, Sonnet 5, and Haiku 4.5, plus legacy Claude 4.x models.
- **Perplexity** — the Sonar family, with an optional **"Sources" section** built from real web citations.
- **Google Gemini** — Gemini 3.6/3.5 Flash for text, and the **Nano Banana** image family (Nano Banana 2, Nano Banana Pro) for stunning illustrations.

Works with the Block Editor (Gutenberg) and the Classic editor, on posts and pages.

= ✍️ Writing, your way =

- Set the number of section headings, and choose whether to include an introduction and a conclusion — the post follows your structure exactly.
- Pick a **default provider and model** once in Settings → AI Blog Post, then override either per post right from the editor sidebar.
- If a post ever hits your token limit, you get a clear notice instead of a silently cut-off article.

= 🖼️ Images that behave =

- Generate up to 4 images per post with Google Nano Banana or OpenAI.
- Images are **saved to your Media Library** with proper alt text — no expiring external URLs, no broken posts a day later.
- **Choose where images go:** spread between sections, at the beginning, after the first paragraph, **after each heading**, at the end — or **pick the exact paragraph for each image** with the custom placement mode.
- One click sets the first image as the **featured image**, correctly synced with the editor.

= 🚀 Future-proof models =

Model dropdowns are curated (latest + still-supported legacy models), and every provider also has a **Custom Model ID** field — when a provider ships a brand-new model tomorrow, paste its ID and use it immediately. No plugin update required.

= 🛠️ Built for developers =

Clean provider abstraction with filters for everything: `martincv_openai_post_providers` (register your own provider), `martincv_openai_post_models`, `martincv_openai_post_system_prompt`, `martincv_openai_post_user_prompt`, `martincv_openai_post_image_prompt`, `martincv_openai_post_request_timeout`, and `martincv_openai_post_post_types`.

== Installation ==

= Installing: =

1. Download the package.
2. Extract it to the "plugins" folder of your WordPress directory.
3. In the Admin Panel, go to "Plugins" and activate it.
4. Go to Settings → AI Blog Post, add the API key(s) for the provider(s) you want, and pick your defaults.

= Upgrading from 1.x: =

Your existing OpenAI API key and settings are migrated automatically the first time you visit the admin after updating.

= Uninstalling: =

1. In the Admin Panel, go to "Plugins", deactivate the plugin and click "Delete". All plugin settings are removed on delete.

= Usage: =

Create a Post or Page and enter a title — the title is the topic. In the sidebar "AI Blog Post" box pick a provider and model (or keep your defaults), set the structure and image options, and click **Generate Post**. The content is inserted into the editor as regular blocks (or classic-editor content), and generated images land in your Media Library.

If you need any additional help feel free to contact me at [martin@martincv.com](mailto:martin@martincv.com).

== Frequently Asked Questions ==

= Which API keys do I need? =

Only the key(s) for the provider(s) you want to use. Each key field in Settings → AI Blog Post links directly to that provider's key page.

= I want to use a model that is not in the dropdown. =

Enter its exact model ID in the "Custom Model ID" field — it overrides the dropdown. This keeps the plugin working with models released after this version.

= Can I control where the generated images are placed? =

Yes. Choose a placement in the editor sidebar: spread between sections, at the beginning, after the first paragraph, after each heading, at the end — or "Custom", which lets you type the exact paragraph number for each image.

= The generation request times out. Why? =

A full post is a single AI request that can take a minute or two depending on the provider, model, and length. If your host enforces a short PHP/proxy timeout, lower Max Tokens or use the `martincv_openai_post_request_timeout` filter.

= The text gets cut off before finishing. =

The plugin will tell you when that happens — increase Max Tokens in Settings → AI Blog Post (default 2048).

== Screenshots ==

1. The Settings screen for the plugin.

2. The AI options in the Add / Edit post screen.

3. Generated Post

== Changelog ==

= 2.0.0 =
_Release Date - 30 July 2026_

**Major update: one plugin, four AI providers**

- New providers: **Anthropic Claude**, **Perplexity**, and **Google Gemini** join OpenAI.
- Pick a default provider and model in Settings; override both per post in the editor sidebar.
- Curated model lists with the latest models (GPT-5.6, Claude Opus 5, Gemini 3.6, Sonar) plus still-supported legacy favorites (GPT-4o, GPT-4.1, GPT-3.5 Turbo, Claude 4.x), and a **Custom Model ID** field for anything newer.
- **Image generation via Google Nano Banana** (Gemini image models) or OpenAI GPT Image 2 — images are saved to the Media Library with alt text instead of hotlinking expiring URLs.
- **Image placement control**: spread between sections, at the beginning, after the first paragraph, after each heading, at the end, or a custom per-image paragraph position.
- Option to set the first generated image as the **featured image**, properly synced with both editors.
- **Perplexity citations**: optionally append a "Sources" section with real links.
- **Truncation detection**: a clear notice when a post hits the Max Tokens limit; migrated 1.x limits are raised to a safe minimum.
- Faster and more reliable: one AI request per post (previously three), cleaner content via native block conversion.
- Fixed from 1.1: generated text was always empty (response parsed with the legacy completions shape); the Max Length setting was never applied; zero values (e.g. temperature 0) reset to defaults.
- Security hardening: capability checks on generation, strict input sanitization, `manage_options` for the settings page, real HTTP status codes on errors.
- Developer filters for providers, models, prompts, timeouts, and post types; settings are cleaned up on uninstall.

= 1.1 =

- Switched to gpt-3.5-turbo via Chat Completions.

= 1.0 =
_Release Date - 14 December 2022_

- Public release of 'MartinCV OpenAi Post'