Add modular browser-based AI threat message detection dashboard

[codezji]

Motivation

• Provide a self-contained browser prototype that simulates a realistic AI moderation engine for detecting threat intent with explainability for hackathon/demo use cases.
• Prioritize layered reasoning and transparent outputs instead of simple keyword matches to make alerts believable and reviewable by humans.

Description

• Add a responsive dashboard UI in index.html with panels for input, results, explainability, recent scans, sample messages, and a status badge.
• Implement a cyber-security themed stylesheet in style.css with dark palette, accent colors, risk meter, and responsive layout.
• Add analysisEngine.js implementing a multi-layer scoring engine (keyword signals, context analysis, behavioral indicators, safety offsets, aggregation, classification, and confidence calculation) with detailed indicator objects for explainability.
• Add uiController.js and main.js to wire UI updates, render explainability and history, support debounced real-time analysis, sample scenarios, and a simulated 1s AI processing delay.

Testing

• Run a JavaScript syntax check using `node -e

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c12f0d26e8

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Include layer scores in empty analysis result

When the input is blank, analyze() returns an object without layers, but UIController.renderResult() always reads result.layers.keywordScore; clicking Analyze Message on an empty/whitespace textarea therefore throws a runtime TypeError and breaks rendering for that run. Keep the return shape consistent (or guard the UI read) so empty input does not crash the dashboard.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Scope keyword regexes to whole words

This alternation applies word boundaries only to the first and last branch, so tokens like bomb can match inside benign words (for example, bombastic), which inflates risk scoring and creates false alerts. The same pattern shape appears in several signal regexes, so wrapping alternatives (e.g., \b(?:...)\b) is needed to preserve detection accuracy.

Useful? React with 👍 / 👎.