Where settings live
In your WordPress admin: Settings → ReplyMind. Settings are split across three tabs — Settings, Logs, and Batch Process. This page documents the Settings tab.
API Key
Single text field. Stores the API key for your selected provider.
| Detail | Value |
|---|---|
| Type | Password input |
| Default | Empty |
| Storage | Option replymind_api_key |
Saving behaviour: Submitting the form with the field empty keeps the existing key. To replace the key, paste a new one and save. To clear the key, click the Remove key button next to the field — it's a separate, nonce-protected action.
The key is never written back into the page DOM. It is readable from the database by users with the manage_options capability.
AI Provider
Choose which API the plugin calls.
| Value | Endpoint | Note |
|---|---|---|
openai (default) |
https://api.openai.com/v1/chat/completions |
Use for sk-... keys |
claude |
https://api.anthropic.com/v1/messages |
Use for sk-ant-... keys |
If your key doesn't match the provider, the plugin shows a mismatch warning on the Settings screen.
Model
Which model to call within the selected provider. The dropdown filters automatically based on Provider.
OpenAI models
| Value | Notes |
|---|---|
gpt-4o-mini (default) |
Best value for most blogs |
gpt-3.5-turbo |
Cheapest, lower quality |
gpt-4o |
Higher quality, ~10× cost vs Mini |
gpt-4-turbo |
Older, comparable to GPT-4o |
gpt-4 |
Most powerful, most expensive |
Anthropic Claude models
| Value | Notes |
|---|---|
claude-haiku-4-5-20251001 |
Fastest, cheapest |
claude-sonnet-4-6 (default) |
Recommended balance |
claude-opus-4-6 |
Most powerful, most expensive |
Enable Plugin
A single checkbox. When unchecked the plugin still loads but does not generate replies for new comments. Default: unchecked.
| Value | Behaviour |
|---|---|
1 (checked) |
New comments are queued for AI reply |
0 (unchecked, default) |
New comments are ignored |
Reply Mode
Controls what happens after a draft is generated.
| Value | Behaviour |
|---|---|
suggestion (default) |
Save the AI reply as a draft. You review and publish manually. |
auto |
Post the AI reply automatically when generated. |
In suggestion mode the AI Reply column shows Draft Ready with an inline preview. In auto mode it shows Published.
Reply Tone
5 options. Sets the tone instruction sent to the AI.
| Value | Default tone instruction |
|---|---|
professional (default) |
Use a professional and respectful tone. |
friendly |
Use a friendly and welcoming tone. |
casual |
Use a casual and approachable tone. |
empathetic |
Use an empathetic and understanding tone, acknowledging the commenter's perspective and feelings. |
formal |
Use a formal and authoritative tone suitable for academic or corporate audiences. |
Response Length
3 options. Controls the AI's length instruction.
| Value | Length instruction |
|---|---|
short (default) |
2–3 sentences maximum. |
medium |
One paragraph of 4–5 sentences. |
detailed |
2–3 paragraphs. |
Reply Language
11 options. Controls what language the AI replies in.
| Value | Behaviour |
|---|---|
auto (default) |
Detect the commenter's language and reply in the same one |
en |
Always reply in English |
es |
Always reply in Spanish |
fr |
Always reply in French |
de |
Always reply in German |
it |
Always reply in Italian |
pt |
Always reply in Portuguese |
hi |
Always reply in Hindi |
ar |
Always reply in Arabic |
ja |
Always reply in Japanese |
zh |
Always reply in Chinese |
System Prompt
Free-form textarea. Included with every AI generation as the system message.
| Detail | Value |
|---|---|
| Type | Textarea (5 rows) |
| Default | Reply professionally and helpfully to blog comments. |
| Sanitization | sanitize_textarea_field |
| Storage | replymind_settings['prompt'] |
The system prompt is combined with the tone instruction, response length instruction, language instruction, post context, and prompt-injection guard before being sent to the AI.
Hidden constants and filters
Some behaviour isn't exposed in the UI but can be changed in code:
Daily reply cap
add_filter( 'replymind_daily_limit', fn() => 500 );
// Set to 0 to disable the cap entirely.
| Detail | Value |
|---|---|
| Default cap | 100 successful API calls per day |
| Counter storage | Transient replymind_daily_YYYYMMDD |
| Reset | Daily, server timezone |
Skip specific comments
add_filter( 'replymind_send_comment_data', function ( $allow, $comment ) {
if ( /* your condition */ ) {
return false;
}
return $allow;
}, 10, 2 );
See Developer reference for the full list of filters and hooks.
All settings at a glance
| Setting | Default | Allowed |
|---|---|---|
| API Key | empty | any string |
| Provider | openai |
openai, claude |
| Model | gpt-4o-mini |
one of the model lists above |
| Enable Plugin | off | on, off |
| Reply Mode | suggestion |
suggestion, auto |
| Tone | professional |
5 tones |
| Response Length | short |
short, medium, detailed |
| Language | auto |
11 codes |
| System Prompt | translated default | any textarea content |