15 versions

AI Engine Pro history.

Every release of AI Engine Pro is archived here with its changelog, file size, and security scan result. Use the archive to roll back to a stable release or audit what changed between updates.

15 of 15 releases scanned clean

v3.6.2Latest

July 16, 20261.9 MBScanned

Fixed

Fixed block editor paragraph spacing broken by the AI Copilot handler wrapper.
Fixed the Anthropic engine dropping conversation history on function-call follow-up requests.
Fixed duplicate embedding vectors for the same reference and environment, which left embeddings stuck at pending after a sync.

Added

Note filtering support for the wp_get_comments MCP tool, allowing agents to read WordPress 6.9 editor Notes.

v3.6.1

July 13, 20261.9 MBScanned

Fixed

MCP user tools not enforcing proper capabilities, allowing Multisite site administrators to edit users beyond their reach. Hotfix. (Thanks to Charles Vosburgh for reporting.)
MCP read tools corrupting block-JSON Unicode escapes by double-unslashing database content.

Improved

Native browser dialogs (alert/confirm/prompt) are now forbidden in code; NekoModal is used instead.

v3.6.0Unavailable

July 11, 20261.9 MBScanned

Added

GPT-5.6 family (Sol, Terra, and Luna) to OpenAI models, with Sol marked as the latest.

Fixed

Gemini image models not being selectable when their ID lacks a `-preview` suffix.
MCP media upload tool crashing on download errors and rejecting extensionless image URLs.

v3.5.9

July 6, 20261.9 MBScanned

Fixed

MCP post meta tools double-serializing array values, which stored nested arrays as strings.
MCP wp_create_post and wp_update_post tools stripping shortcodes, email HTML, and inline CSS for admin-authorized callers.
Chatbot theme styles loading outside AI Engine's admin screens, which caused Gutenberg editor iframe warnings.
simpleJsonQuery failing on Anthropic and Google environments by no longer forcing an OpenAI fallback model, with a clearer error message.
Discussions and Queries timestamps now display in the site's configured timezone instead of the browser's.
MCP Functions settings crashing when tools were registered due to the list being returned as an object instead of an array.
OpenAI Vector Store Pull All failing with a limit error by paginating the file list.

Improved

WordPress MCP tools are now enabled by default.

v3.5.8

July 3, 20261.9 MBScanned

Fixed

Code blocks now display correctly in the Timeless and Foundation chatbot themes (was using proportional font instead of monospace).
Sonnet 5 breaking with "each thinking block must contain thinking" when streaming with function calling.
MacOS chatbot container showing empty notches in the corners due to double-rounded borders.

Improved

Reorganized the admin UI — Modules tab now leads with "Chatbots & Knowledge" and "MCP & Agents", and Settings tabs were reordered to match.
Moved the Event Logs option from Settings → AI to Settings → Chatbot for easier discovery.
Enlarged the chatbot preview pane in settings (480px → 620px) and polished the ChatGPT theme with monospace code blocks, accent border on focused input, and smoother transitions.
Refined the Messages chatbot theme with monospace code blocks, readable links in user bubbles, input focus ring, and capped bubble width.

Added

Gutenberg MCP tools: wp_write_blocks, wp_list_block_patterns, and wp_insert_block_pattern.
Nonce checks added to license, rating, and news notice forms.

v3.5.7Unavailable

June 30, 20262.0 MBScanned

Fixed

Chroma "Pull All" now returns all pages of large collections instead of only the first page.
Anthropic function-calling loops no longer fail when no explicit max tokens is set.
Gemini Interactions engine now properly includes RAG context, chat history, and PDF uploads, and passes connection tests.
Outbound API calls now verify TLS certificates by default.

Added

Claude Sonnet 5 model support.
In-chat image output and stream error handling for Gemini Interactions.

Improved

Credited Shivamani Vastrala and Meher Sudhakar Abbireddi (via WPScan) for reporting the security issues fixed in 3.5.5.

v3.5.6Unavailable

June 26, 20261.9 MBScanned

Added

Gemini Interactions API engine as the default for Google, with streaming, function calling, and in-chat image generation (classic API kept as a fallback).
Google Search and Maps grounding as Gemini chatbot tools, plus a setting to fall back to the standard Gemini API.
MCP server support to Realtime voice chatbots — tools were selectable but previously did nothing — and the model now speaks tool results aloud.
`mwai_mcp_mutate` action that fires after any content-mutating MCP tool, letting integrations purge caches, reindex, or audit.
`raw` option to the MCP `wp_get_option` tool to read directly from the database, bypassing object cache and filters.
`author__not_in` to MCP `wp_get_posts`, and a `content_format=prose` option to `wp_get_post`/`wp_get_post_snapshot` that strips block-attribute JSON to reduce token usage.

Fixed

`wp_create_post` MCP tool no longer corrupts shortcode and HTML content via the markdown parser.
MCP log entries mislabeled as OAuth; admin and bearer calls now display correctly.
Streaming parallel tool calls leaving a null in the `tool_calls` array, which broke function calling on strict providers.
MCP `wp_get_posts`, `wp_get_media`, and `wp_get_comments` filters by author, `user_id`, and email were accepted but silently ignored.
`wp_update_option` now decodes JSON-string array values into real arrays before saving, preventing corruption of array options like `sticky_posts`.
Chatbot events panel is now scrollable and error messages are dismissable.

Improved

Removed the legacy MCP SSE transport, keeping only the Streamable HTTP endpoint.
Enabled function calling for Mistral.
Uploaded images now fill the message bubble width in the chatbot across all themes.
MCP post tools now accept `ID`, `post_id`, or `id` interchangeably, and `wp_update_post` accepts top-level fields with clearer error messages.

v3.5.5Unavailable

June 20, 20261.9 MBScanned

Fixed

Path traversal vulnerability in create_image and prevented chatbot discussions from being taken over by another logged-in user.
Add-ons were reported as disabled whenever options were fetched outside the admin context.
Max Tokens entered with a thousands separator (e.g. the Danish "64.000") was read as 64 and capped responses.
Content Aware placeholders now resolve automatically in chatbot instructions without requiring a toggle.
Setup Assistant no longer shows "Connect an AI Provider" as incomplete when only the Fast model is deprecated.
Realtime instructions now resolve placeholders like {FIRST_NAME} into their values.
MCP tools generated from WordPress REST endpoints could break Google Gemini due to enum on non-string types.
Changing the type of an AI environment now drops fields from the previous type (endpoint, region, deployments), preventing engines from calling the wrong provider's endpoint.
The "Default AI Environment not configured" warning no longer appears when a Chroma embeddings environment generates its own vectors.
The Usage widget on the Dashboard no longer overflows its panel when numbers are wide.
Anthropic's streaming ping keepalive events are now ignored instead of being logged as an unknown stream data type.

Improved

Reworked embeddings sync — pages are no longer marked stale without real changes, the background task (now called Maintenance) runs for all environments, processes batches within a time budget, no longer gets blocked by a failing vector, and can be disabled in Settings > Knowledge > Sync Settings.
API key and token fields are now masked with an eye toggle to reveal them.
Marked the original GPT-5 models (gpt-5, gpt-5-mini, gpt-5-nano, gpt-5-pro) as deprecated per OpenAI's December 10, 2026 shutdown.
The fullscreen resize button now uses thin corner-bracket icons that match the close button.
OpenAI Vector Store document uploads now accept multiple files at once.

Added

User Memory added to the add-ons list.
Optional capability filter for get_chatbots() to fetch only chatbots whose model supports function calling.

Other

🎵 Discuss with others about Ai Engine on [the Discord](https://discord.gg/bHDGh38).
🌴 Keep us motivated with [a little review here](https://wordpress.org/support/plugin/ai-engine/reviews/). Thank you!
🥰 If you want to help us, check our [Patreon](https://www.patreon.com/meowapps). Thank you!
🚀 [Click here](https://trello.com/b/8U9SdiMy/ai-engine-feature-requests) to vote for the features you want the most.

v3.5.4Unavailable

June 10, 20261.9 MBScanned

Fixed

Hold to Talk button on mobile Realtime chatbot no longer triggers the context menu (copy, select) when held.
Sending is now blocked while a file is still uploading, with a spinner on the send button and a progress bar on the file preview until upload completes.
Eliminated the gap between the chatbot and the keyboard on iOS by locking body scroll while the mobile chatbot is open.
Anthropic safety refusals now show a readable message instead of an empty reply.
OpenClaw link in the readme now points to our own article.
PHP 8.2+ deprecation warning caused by an undeclared `$extraData` property on `Meow_MWAI_Reply`.
`wp_alter_post` regex mode now handles patterns containing "/" (e.g. Gutenberg block markers) and returns the actual PCRE error instead of a generic one.

Improved

Images and files can now be sent without a text message; engines skip the empty text part for providers that reject it.
AI provider connectors now mentioned in the readme description.
PDF import now remembers the Density, Overlap, and filename-as-title settings between imports.
Marked GPT Image 1.5 and GPT Image 1 Mini as deprecated; default image model moved to GPT Image 2.

Added

Claude Fable 5 model. Marked Claude Sonnet 4, Opus 4, and Opus 4.1 as deprecated.
Web search support for OpenRouter via the Tools & Capabilities section.
Full Screen view for Discussions, a Discussions count on the Dashboard, and fixed filenames with underscores being rendered as italics in the admin discussion view.
Embeddings sync can now be restricted to specific Polylang languages via a comma-separated languages field.

v3.5.3Unavailable

May 29, 20261.8 MBScanned

Fixed

Filenames with underscores (e.g. `Some_File_2026.pdf`) were getting partially italicized by the chatbot's markdown renderer.
ReferenceError (`data is not defined`) in the chatbot response handler that masked the real error message when a host returned a non-JSON body.
The Push All button showed "(null)" when a post type had zero matching posts.
The `wp_update_option` MCP tool schema no longer logs a "problematic union type" warning on every `list_tools` call.
The Meow Apps menu icon now shows correctly in both the expanded and collapsed sidebar states.
Hotfix for the MCP endpoint URL being hidden until a bearer token was generated, even though OAuth doesn't need one. It now appears as soon as the MCP Server is enabled.

Added

Claude Opus 4.8 to the Anthropic model list, with the `latest` tag moved to it.
AI Forms multi-upload fields now stack new file selections on top of previous ones, with a staged list and per-file remove buttons.
`mwai-bouncing-loader` and `mwai-bouncing-dot` CSS classes to the chatbot loading dots for custom styling.
MCP Logs view in Insights with per-tool-call capture, status, duration, and a top tools widget.
MCP Connection Test in MCP Server settings that detects host-level WAF blocks (e.g. WP Engine's default) before they surface in claude.ai.

Improved

Aligned Forms and Playground markdown rendering with the chatbot so single newlines now produce visible line breaks.
Extended CORS headers on MCP endpoints with Mcp-Protocol-Version, Mcp-Session-Id, Accept, and Last-Event-ID for browser-context MCP clients.
Claude Sonnet 4 and Opus 4 (2025/05/14) marked as deprecated ahead of Anthropic's 2026-06-15 retirement date.
The chatbot Functions empty state now tells Free users that function calling is Pro-only.
Polished the plugin readme description with cleaner punctuation and brand consistency.

v3.5.2Unavailable

May 22, 20261.8 MBScanned

Added

Setup Assistant on the Dashboard that walks through essential configuration with a progress bar and per-step actions.
Direct file upload to OpenAI Vector Store via a new Documents tab, with async status polling, Sync from OpenAI action, and clearer PHP upload-limit errors.
`setField` and `prefill` methods to the AI Forms JS API for setting values programmatically without manual `dispatchEvent` workarounds.

Improved

Redesigned the Dashboard Usage panel into an analytics card with headline, week-over-week delta, daily bar chart, and provider/peak/average/per-query tiles.
Reorganized the Modules tab into clearer sections with short intros, concise descriptions, renamed Utilities to AI Copilot, and surfaced MCP and Public API as proper modules.
Renamed the user-facing "Units" metric to "Tokens" across the dashboard, queries explorer, insights, and `[mwai_stats]` shortcode, with full back-compat on REST output and legacy settings.
Aligned the Discussions, Insights, and Knowledge layouts by standardizing column ratios and removing custom margin and header overrides.
Added a Refresh button to the Usage panel header, widened the Tokens column in Queries Explorer, and showed "N/A" for queries with no scope.
MCP tool-registration warnings are now always written to the PHP error log instead of requiring the debug logging flag.

Fixed

Stripped inline citation markers from OpenAI's file_search Responses API text that were surfacing as gibberish in chatbot replies.
Stopped embedding the sessionId in chatbot and forms HTML for logged-out visitors, so page caches no longer share one sessionId across all visitors.
Guarded REST AI endpoints so they no longer overwrite the cookie-derived session with an empty client-sent sessionId.
Gemini embedding queries no longer report bogus token counts by correctly reading promptTokenCount from the response.
Dynamic MCP REST tool schemas no longer serialize the meta arg on media/posts/pages as an array, which Claude's MCP validator was rejecting.
Translate Post now translates list items inside modern Gutenberg lists, which were silently skipped since WP 6.1+.

v3.4.8Unavailable

May 10, 20261.8 MBScanned

Fixed

Resolved OpenAI Responses API "No tool output found" errors during chained recursive tool calls and static-only turns. Static-skip is now disabled for Responses API because every `function_call` must be answered.
Managed-mode WP AI Client and AI Engine integrations now satisfy WordPress 7 AI plugin credential requirements, preventing false rejections for Ollama, OpenRouter, Mistral, Perplexity, and Replicate.
Removed redundant vector-DB addon initialization to prevent idle "OpenAI Vector Store ID is not configured" logs.
Resolved PHP 8.4 nullable-parameter deprecations in query and MCP helpers.
ChatML base now supports string-shaped `error` fields in API responses.

Added

Custom OpenAI-compatible provider for Ollama, LM Studio, vLLM, llama.cpp, LocalAI, and similar servers.
Built-in xAI/Grok provider with dynamic model fetching and fallback support.
Insights retention setting with daily cleanup for old log rows. Defaults to Never.
Embedding chunk metadata: source, partIndex, and partTotal, exposed through `mwai_embeddings_vector_metadata` for Pinecone, Qdrant, Chroma, and OpenAI Vector Store. PDF imports can now use the filename as the chunk title, and Modify Embedding can edit the new fields.
Code Engine static/dynamic Behavior now applies to Callable function snippets, skipping AI feedback when all calls are static.

Improved

Reworked onboarding with a default OpenAI environment, Knowledge provider chooser, one-click OpenAI vector store creation, no-key welcome message, and AI Assistant enabled by default.
Reworked Pro license management with a Modify License flow for Re-Validate, Modify, and Remove actions.
Hardened MCP handling by releasing PHP session locks, using proper trash hooks, deduplicating cache busts, enforcing `confirm_write` for non-read SQL, and purging LiteSpeed/WP Rocket caches after post writes.
Usage widget preferences for Price/Units/Queries and Daily/Monthly now persist across reloads.
Query and Reply tabs now point users to Settings > Others > Insights when data is unavailable.
Removed the legacy snippet-vault fallback and unused per-environment usage field.

Why archive

Sometimes the latest release isn't the one you need. The archive lets you pin AI Engine Pro to a known-good version, or roll back while a bug is investigated. Every release stays scanned and reachable.