Qgenius AI Learning Terminal

AI Learning Terminal is a WordPress plugin for embedding AI-simulated learning terminals in lessons, posts and member pages.

It does not execute code. It sends learner input to a configured AI provider and asks the model to simulate likely output, explain mistakes and provide teaching feedback.

Features

• Terminal Profile management
• Shortcode terminal embedding
• REST API endpoint for simulation
• OpenAI, Anthropic, DeepSeek, OpenRouter, Ollama and Custom API settings
• Profile-level model overrides
• Rate limits and guest controls
• Full, stats-only or disabled logging
• AI response schema normalization
• CodeMirror progressive enhancement with textarea fallback
• Learning context fields and completion checks
• User practice history shortcode
• Profile import/export with conflict strategies
• Bundled example profiles and JSON schema

Install

1. Copy qgenius-ai-learning-terminal into wp-content/plugins.
2. Activate Qgenius AI Learning Terminal in WordPress.
3. Open AI Learning Terminal > Settings.
4. Configure a provider API key and default model.
5. Open AI Learning Terminal > All Terminals and copy a shortcode.

Provider Setup

Global provider settings live in AI Learning Terminal > Settings.

Supported providers:

• OpenAI
• Anthropic
• DeepSeek
• OpenRouter
• Ollama
• Custom API

For production sites, API credentials can be supplied from wp-config.php:

define('QAILT_OPENAI_API_KEY', '...');
define('QAILT_ANTHROPIC_API_KEY', '...');
define('QAILT_DEEPSEEK_API_KEY', '...');
define('QAILT_OPENROUTER_API_KEY', '...');
define('QAILT_CUSTOM_API_KEY', '...');
define('QAILT_CUSTOM_API_ENDPOINT', 'https://example.com/v1/chat/completions');
define('QAILT_OLLAMA_ENDPOINT', 'http://127.0.0.1:11434');

Constants override database-stored settings.

Terminal Shortcode

[ai_learning_terminal id="python_beginner" height="500" theme="dark"]

Supported attributes:

• id required Terminal Profile ID
• height editor minimum height
• theme dark or light
• show_title true or false
• default_code
• exercise_id
• lesson_id

Learner History Shortcode

[ai_learning_terminal_history limit="20" terminal_id="python_beginner" show_input="true" show_response="false" allow_clear="true"]

The history shortcode only shows the current logged-in user's records. When allow_clear is true, logged-in users can delete their own displayed practice history. A terminal_id limits clearing to that terminal.

Terminal Profiles

Profiles define the learning behavior of a terminal:

• Terminal name and ID
• Terminal type
• Provider/model overrides
• Learner level
• Feedback style
• Enabled actions
• System prompt
• Learning goal
• Task brief
• Evaluation criteria
• Course context
• Completion policy
• Examples and reference answer

Enabled actions:

• simulate_run
• debug
• explain
• hint
• next_challenge
• check_completion

Profile Import and Export

Profiles can be exported one at a time from All Terminals, or exported in bulk from Import / Export.

Import supports conflict strategies:

• Overwrite existing profile
• Skip existing profile
• Import as a copy

Bundled example profiles live in profiles/:

• python-beginner.json
• sql-beginner.json
• skills-coach.json
• prompt-coach.json
• agent-workflow-coach.json
• git-beginner.json
• shell-safety-coach.json

The profile schema is available at profiles/profile.schema.json.

Security and Cost Controls

Global settings support:

• Guest access on/off
• Guest and logged-in daily request limits
• Per-IP per-minute request limits
• Maximum input length
• Log mode: full logs, stats only, or no logging
• Automatic daily pruning by retention days
• No-cost connection checks where providers expose a model/list endpoint

See SECURITY.md and PRIVACY.md for details.

Development

Run local checks:

sh scripts/validate.sh

See docs/DEVELOPMENT.md for architecture notes.

License

GPLv2 or later. See LICENSE.