| 开发者 | maruhidedev |
|---|---|
| 更新时间 | 2026年10月30日 17:19 |
| PHP版本: | 7.4 及以上 |
| WordPress版本: | 7.1 |
| 版权: | GPLv2 or later |
| 版权网址: | 版权信息 |
/wp-content/plugins/ and activate it, or install it from the Plugins screen.Yes. Every feature — manual and daily generation, auto-retry, missed-publish recovery, scheduled publishing, and the eyecatch generator — is free to use. There is no paid tier, license key or usage limit in the plugin. You only pay your own AI provider for generation.
Japanese only. The built-in prompt is written for Japanese articles (about 2,000+ characters with H2/H3 headings). Other languages are not supported yet.
No. Your key is stored only in your site's database and is sent only to the AI provider you configured (Anthropic or OpenAI) at generation time.
In this plugin's settings row in your site database (wp_options), without encryption. Anyone with database or administrator access to your site can read it, so keep database backups private. If you use "WordPress AI", the key is managed by WordPress core in Settings → Connectors instead.
By default, "draft / review mode" is on, so articles are saved as drafts for a person to review before publishing. You can turn on daily generation and scheduled publishing in the settings.
Any current model name accepted by your provider's API, entered in the settings. The default is claude-sonnet-5-5. Providers retire old models from time to time; if the log says the model name was not found, change it to a current model.
Every hour it looks for posts still marked "Scheduled" (status future) whose publish time has passed, and publishes them with wp_publish_post(). By default this covers all standard posts on the site, as in earlier versions. In the settings you can limit it to posts created by this plugin (posts created before 0.6.0 are not marked and will then be skipped).
Daily generation and the hourly recovery run on WP-Cron events. If you disable WP-Cron, run wp-cron.php from a real server cron job. If Action Scheduler is installed, generation jobs use it automatically.
It depends on the model and your provider's pricing. Each request allows up to 16,000 output tokens; in the developer's tests a Japanese article used roughly 2,700–5,400 output tokens. The plugin itself is free.
Check the generation log on the settings page. Common reasons: the API key or billing was not accepted, the model name is not available, or the AI stopped at its output limit (the incomplete article is not saved).
max_completion_tokens.claude-sonnet-5-5 for new installs (existing settings are kept).