PressHub AI Co-Pilot

Enterprise AI Newsroom, Editorial Co-Pilot & Multimedia Publishing Pipeline for WordPress.

🌟 Overview

PressHub AI Co-Pilot transforms WordPress into an autonomous, AI-driven newsroom and multimedia production suite. Spanning from multi-source web/RSS news gathering to editorial quality scoring, bilingual daily briefings, multi-host podcast script generation, and broadcast-quality text-to-speech synthesis, PressHub AI provides news organizations and content creators with an enterprise-grade automated editorial workflow.

Built with a modular Provider Store (supporting OpenAI, Anthropic, Google Gemini, and Google Cloud Vertex AI/TTS), 4-layer enterprise observability, and strict architectural principles (Settings-First, Zero-Silent-Fallback, Clean-Slate), PressHub AI ensures total transparency, deterministic control, and reliability.

🚀 Core Subsystems & Features

┌────────────────────────────────────────────────────────────────────────┐
│                        PressHub AI Newsroom                            │
└──────────────────────────────────┬─────────────────────────────────────┘
                                   │
      ┌────────────────────────────┼────────────────────────────┐
      ▼                            ▼                            ▼
┌──────────────────┐     ┌──────────────────┐     ┌──────────────────┐
│  Daily Briefing  │     │ Podcast Producer │     │  In-Editor AI    │
│     Pipeline     │     │   & TTS Studio   │     │     Co-Pilot     │
├──────────────────┤     ├──────────────────┤     ├──────────────────┤
│ • Morning & Noon │     │ • Multi-Host Dialogue│ │ • Draft Assistant│
│ • RSS & Web Scrape│    │ • Nameless Hosts │     │ • 0-100 Scorecard│
│ • Context Ingest │     │ • Spoken Dates   │     │ • Chat Intents   │
│ • Image Cycling  │     │ • Audio-Only Post│     │ • Imagen Sideload│
│ • Social Meta    │     │ • Script Editor  │     │ • Block Revision │
│ • Stage Triggers │     │ • Audio Re-Gen   │     │ • Revision Logs  │
└────────┬─────────┘     └────────┬─────────┘     └────────┬─────────┘
         │                        │                        │
         └────────────────────────┼────────────────────────┘
                                  ▼
┌────────────────────────────────────────────────────────────────────────┐
│                  Decoupled Provider Store & Engine                     │
│  • OpenAI    • Anthropic Claude    • Google Gemini    • Google Cloud   │
├────────────────────────────────────────────────────────────────────────┤
│                  4-Layer Observability & CLI Diagnostics               │
│  Layer 1: Structured  │ Layer 2: Token Log │ Layer 3: Audit │ L4: Core │
└────────────────────────────────────────────────────────────────────────┘

1. 📰 Automated Daily Briefing Pipeline (Morning Edition & Noon News Update)

Autonomous News Harvester (PressHub_AI_News_Harvester): Concurrently scrapes and processes articles from configured RSS feeds and direct web URLs. Features domain politeness delays, per-domain concurrency limits, total execution time budgets, lead-article extraction, and HTML sanitization. Manages edition-partitioned raw snapshots (raw-articles.json vs raw-articles-noon.json).
Intelligent News Curator (PressHub_AI_News_Curator): Evaluates harvested candidate stories using LLM analysis. Employs semantic article deduplication to prevent repetitive coverage, enforces strict token context limits, and produces comprehensive bilingual (Greek and English) daily briefings.
- Noon News Update: Re-scrapes sources at noon, automatically ingests the morning briefing post as reference context, and generates an updated noon briefing focusing on developments since morning.
Featured Image Pools & Interactive Media Library Picker: Configure dedicated pools of Media Library images for Morning briefings (presshub_ai_briefing_featured_image_ids_morning) and Noon updates (presshub_ai_briefing_featured_image_ids_noon), with automatic fallback. Features an interactive WordPress Media Library picker UI with multi-select modal, thumbnail preview cards, badge IDs, and individual remove/clear actions. Images are cycled at random and assigned as post thumbnails to daily text briefings and podcasts, with intelligent same-day collision avoidance preventing duplicate featured images between morning/noon text briefings and podcasts when alternative images exist in the pool.
Social Sharing & OpenGraph Optimization (PressHub_AI_Social_Meta): Emits rich, opengraph.xyz-compliant metadata (og:title, og:description, og:image, og:site_name, og:type=article, twitter:card=summary_large_image, twitter:title, twitter:description, twitter:image, and standard <meta name="description">). Includes an automated title length optimizer guaranteeing document titles remain ≤ 60 characters for optimal search snippet display.
Interactive Briefing Admin (PressHub_AI_Briefing_Admin): A dedicated WordPress administrative control center providing visibility into each pipeline stage (Harvest → Curate → Audio/Podcast → Publish), stage event timelines, manual triggers, single-stage regeneration, and full timezone-aware scheduling. Includes interactive tabs to toggle between Morning Edition and Noon Update, dynamic in-browser audio player mounting on synthesis completion, and role-gating strictly for Editors and above (edit_others_posts, filterable via presshub_ai_briefing_cap).

2. 🎙️ Podcast Producer, Script Editor & Audio Regeneration

Multi-Host Script Engine (PressHub_AI_Podcast_Producer): Automatically transforms curated news briefings into dynamic, conversational podcast scripts between multiple hosts:
- Nameless Speaker Invariant: Presenters never awkwardly refer to each other or themselves by name; dialogue progresses naturally through ideas.
- Spoken Date Formatting: All dates are converted to natural spoken-language phrasing (e.g. "3rd of May 2026" or "3 Μαΐου 2026").
Configurable Title Prefixes: Separate editorial title prefix options for Morning Podcast (presshub_ai_briefing_podcast_title_prefix, default "Πρωινό Podcast Ενημέρωσης - ") and Noon Podcast (presshub_ai_briefing_noon_podcast_title_prefix, default "Μεσημβρινό Podcast Ενημέρωσης - "), configurable from Settings.
Clean Audio Post Content & Front-Page Excerpt Suppression: By default, podcast post content displays exclusively the audio player block (wp:audio), and raw dialogue text is suppressed from post_excerpt and filtered out of front-page post previews, loop cards, and public RSS feeds (configurable via presshub_ai_briefing_podcast_include_text). Social meta tags emit a clean editorial audio description instead of leaking dialogue turns.
Podcast Dialogue Editor Metabox & One-Click Audio Regeneration: Editors can review and refine spoken scripts in the "Podcast Audio & Dialogue Editor" metabox on podcast edit screens, and click "Regenerate Audio from Script" to re-synthesize and stitch the episode in real time.
Resilient Audio Synthesis & Cron Error Isolation: Synthesizer accepts nullable custom scripts and automatically shifts parameters (e.g. synthesize_podcast($date, 'noon')), while briefing crons encapsulate generation and synthesis in try/catch blocks with structured logging to eliminate unhandled runtime fatal errors.
Broadcast Audio Synthesizer & Gemini 3.8 TTS REST Architecture (PressHub_AI_Audio_Synthesizer & PressHub_AI_API_Client):
- Zero Narrated Instructions: Passes clean verbatim dialogue turns in contents[0].parts[].text without prepended prompt directives, eliminating prompt narration on modern models (gemini-3.8-flash-tts / gemini-3.8-flash-lite-tts).
- Official Gemini 3.8 TTS REST Standard (structured — Recommended): Conforms strictly to Google's official Gemini 3.8 TTS REST specification. Turn-level speech_metadata.style carries delivery emotion, speaking rate (0.85x - 1.25x), and voice pitch tuning (-5.0 to +5.0 semitones) without top-level system instructions. Inactive settings (e.g. system instructions) dynamically grey out in the UI with informative badges.
- Hybrid / System Instruction Override Mode (hybrid_instruction): Passes both turn-level speech_metadata.style and top-level systemInstruction with resolved speaker persona and rate placeholders.
- Multi-Speaker Persona Mapping: Maps Presenter 1 and Presenter 2 to distinct voice personas (e.g. Charon, Kore, Fenrir, Aoede) in speechConfig.multiSpeakerVoiceConfig.
- RIFF WAV Header Handling: Detects RIFF headers on incoming audio bytes from Gemini 3.8 to prevent double-wrapping WAV headers.
Gemini Speech API JSON Inspector & Live Voice Testing Studio:
- Live preview container in Daily Briefing settings renders the exact JSON request body sent to Gemini's speech endpoint, updating dynamically as voices, speed, pitch, delivery style, or test scripts change.
- Interactive testing studio allows editors to audition 1-host and 2-host audio synthesis with immediate in-browser audio playback, execution metrics (duration ms, byte size), and download capabilities before publishing. Accessible directly from the Daily Briefing Hub Audio Synthesis Stage Card.

3. ✍️ In-Editor Editorial Co-Pilot

AI Draft Generation: Accelerates content creation by synthesizing comprehensive drafts from raw source URLs, interview transcripts, or bullet notes.
4-Key Editorial Scorecard & Legal Exposure Assessment: Evaluates drafts using a structured 4-key rubric: overall publication readiness score (0–100), legal exposure rating badge (Low, Medium, or High), 2–3 sentence editorial feedback, and targeted legal concern alerts highlighting potential defamation or unverified claims.
Block-Level Gutenberg Editing: Modifies individual Gutenberg block attributes (attributes.content / attributes.text) without resetting document state, supporting existing article revisions.

4. 🖼️ AI Featured Image Generation & Studio Engine

Photojournalistic Editorial Realism: Analyzes article content and generates visual prompts adhering to 35mm documentary photography aesthetics with natural lighting and realistic textures, strictly barring "AI plastic" looks, surreal saturation, or typography/text in the synthesized image.
Configurable Provider & Models: Powered by Google Gemini (gemini-3.1-flash-image via multimodal generateContent) or OpenAI (dall-e-3 via images/generations) with configurable aspect ratios (16:9, 1:1, 4:3). The image model is determined directly by the selected AI provider's configured model in the Provider Store.
Dedicated Settings Tab ("AI Featured Images"): Control prompt analysis templates, image generation providers, and aspect ratios directly from PressHub AI → Settings (#images).
Daily Briefing Hub Selection & Resilient Pool Fallback: Briefings and podcast production can independently select between the Featured Image Pool or Automatic AI Generation, with automatic, graceful fallback to the image pool if synthesis fails or times out.
Post Editor "AI Featured Image Studio": Embedded inside the post editor metabox, allowing journalists to:
- Draft photojournalistic visual prompts from current article content with one click.
- Review and edit the prompt in an editable textarea (persisted in post meta _presshub_ai_image_prompt).
- Synthesize and regenerate images directly into the WordPress Media Library with live preview.
- Assign generated images as the article's official featured image with one click.

5. 🎛️ Instruction Presets & Prompt Studio

Multi-Tier Presets: Admin-curated global defaults, author-specific profiles, author defaults, and per-request metabox dropdown selection. Presets append specialized instructions to base system prompts without overwriting foundational editorial safety constraints.
Externalized Prompt Assets: Prompts are stored as modular text files under presshub-ai-editor/assets/prompts/ (e.g. default_greek_briefing.txt, podcast_style_two_hosts_greek.txt, article_image_analysis_prompt.txt) and dynamically managed via PressHub_AI_Prompt_Loader.
Admin Prompt Studio: Inspect, customize, and reset prompt templates directly within WordPress settings without editing source code.

🏛️ Architectural Principles

PressHub AI adheres strictly to enterprise software engineering standards:

⚙️ Settings-First Principle

No operational limit, threshold, cap, time budget, batch size, retry count, or rate limit that affects user-visible behavior may be hardcoded inside business logic.

Every parameter is registered in the WordPress Options API with strict sanitization bounds, exposed via PressHub_AI_Settings_Storage, and rendered in the admin settings UI. Module-level token limits and HTTP timeouts enforce minimum safe thresholds (e.g. max_tokens ≥ 64, timeout ≥ 10s) ensuring unrendered form submissions or empty inputs never corrupt model execution parameters.

🛡️ Zero-Silent-Fallback Principle

No business logic, API client, prompt resolver, or pipeline stage may silently fall back to a hardcoded model, prompt, or operational parameter when configuration is missing or failing.

Missing keys, unconfigured models, or failed API calls halt immediately with explicit WP_Error objects or descriptive exceptions, logging full diagnostic details to structured logs. Hidden code fallbacks and silent multi-model retry loops are prohibited.

🧹 Clean-Slate Architecture

The Provider Store is the sole source of truth. No legacy option shadowing or backward-compatibility baggage.

All provider parameters and default models resolve through PressHub_AI_Provider_Store (wp_presshub_ai_providers). Deprecated options are actively purged via idempotent database migrations rather than lingering in resolution cascades.

🪵 Enterprise Observability & 4-Layer Logging

PressHub AI implements a 4-layer logging architecture to provide total transparency into every automated decision, token cost, and administrative action:

Layer	Purpose	Target / Storage	Inspection Tool
Layer 1: Structured Logs	Subsystem operations, harvesting steps, TTS synthesis, HTTP events	`wp-content/presshub-debug.log`	`php dev-env/scripts/tail-logs.php --source=presshub`
Layer 2: Token Telemetry	Model IDs, prompt/completion tokens, latency, cost, and JSON metadata	`wp_presshub_ai_token_logs`	`php dev-env/scripts/view-token-logs.php`
Layer 3: Security & Audit	Settings updates, provider credential changes, and admin actions	`wp_presshub_ai_audit_log`	`php dev-env/scripts/query-db.php "SELECT * FROM wp_presshub_ai_audit_log"`
Layer 4: WordPress Core	PHP runtime warnings, fatal tracebacks, cURL/environment issues	`wp-content/debug.log`	`php dev-env/scripts/tail-logs.php --source=wp`

CLI Observability Cheat-Sheet

# Stream all PressHub logs in real time
php dev-env/scripts/tail-logs.php --follow

# Filter logs by error level
php dev-env/scripts/tail-logs.php --level=ERROR

# Inspect aggregate LLM token usage, latencies, and provider breakdowns
php dev-env/scripts/view-token-logs.php --stats

# View recent token logs with detailed JSON request/response metadata
php dev-env/scripts/view-token-logs.php --detail=1

# Query the SQLite database directly
php dev-env/scripts/query-db.php "SELECT provider, model, total_tokens, duration_ms FROM wp_presshub_ai_token_logs ORDER BY id DESC LIMIT 5"

⚙️ Settings Reference Guide

Navigate to PressHub AI → Settings to configure the platform across responsive, per-tab AJAX-saving sections:

Providers: Register API keys, organization IDs, and default models for OpenAI, Anthropic, Google Gemini, and Google Cloud (Vertex AI/TTS). Masked secrets prevent accidental credential exposure.
News Harvester: Configure RSS source URLs, fetch timeout budgets, domain polite delays, concurrency limits, and lead article filtering.
News Curation: Set curation LLM models, curation temperature, maximum article candidate pools, character caps, and language presets.
Podcast & Audio Studio: Configure default podcast presenter personas, audio synthesis voices (male/female, accents), speaking rate, pitch modulation, and audio segment stitching.
Prompt Studio: Customize system prompts and editorial guidelines for drafts, research, daily briefings, and podcast scripts.
Rate Limits & Security: Enforce per-user request rate limiting windows, IP safeguards, and automated research-log retention sweeps.
Logs & Maintenance: Configure log retention days, log verbosity (DEBUG, INFO, WARNING, ERROR), prompt debug logging, and token usage analytics.

💻 Local Development & Testing

The repository features a self-contained local WordPress development environment powered by the official WordPress Core SQLite engine (zero external MySQL/MariaDB or Docker requirements).

Environment Layout

presshub/
├── presshub-ai-editor/          # Core WordPress plugin (PHP)
│   ├── includes/                # Business logic, modules, Provider Store
│   ├── assets/                  # Sidebar UI, admin JS/CSS, prompt assets
│   └── tests/                   # 73 isolated PHP unit test suites
├── presshub-workflow/           # TypeScript workflow & MCP server
│   ├── src/                     # MCP scorecard server & sync utilities
│   └── tests/                   # 6 Jest test suites (81 tests)
├── dev-env/                     # Local SQLite WordPress runtime & CLI tools
│   ├── scripts/                 # tail-logs, view-token-logs, query-db, setup
│   └── wordpress/               # Local WordPress installation
└── docs/                        # Specifications, plans, review records

1. Starting the Dev Server

# Start local WordPress instance at http://127.0.0.1:8888
php dev-env/scripts/server.php

Front-end: http://127.0.0.1:8888
WP Admin: http://127.0.0.1:8888/wp-admin/ (User: admin, Password: password123)
Live Code Sync: The plugin directory is dynamically linked via a filesystem junction (mklink /J) directly to presshub-ai-editor/.

2. Running PHP Unit Tests

The PHP test suite runs in isolation without external database dependencies:

cd presshub-ai-editor
php tests/run-all-tests.php

Executes all 73 test suites with 100% pass rate.

3. Running TypeScript & MCP Workflow Tests

The workflow package integrates Model Context Protocol (MCP) servers and editorial automation:

cd presshub-workflow
npm install
npm test
npx tsc --noEmit

Executes all 6 Jest test suites (81 tests) with strict TypeScript type-checking.

4. Running WP-CLI Commands

# Via Windows batch wrapper
dev-env\bin\wp plugin list
dev-env\bin\wp eval "echo get_option('presshub_ai_log_level');"

# Or via PHP
php dev-env/bin/wp-cli.phar --path=dev-env/wordpress <command>

📦 Requirements & Installation

Requirements

WordPress: ≥ 6.0 (tested up to 6.7)
PHP: ≥ 7.4
API Keys: At least one active provider key (OpenAI, Anthropic Claude, or Google Gemini / Google Cloud Vertex AI & TTS).

Installation Steps

Download the latest release zip (presshub-ai-editor-x.x.x.zip) from the GitHub Releases tab.
In your WordPress admin dashboard, navigate to Plugins → Add New → Upload Plugin.
Choose the downloaded zip file and click Install Now.
Activate PressHub AI Co-Pilot.
Navigate to PressHub AI → Settings to configure your provider API keys.

Automatic Updates

The plugin embeds an integrated GitHub update checker. Releases published on the main branch with attached zip assets are automatically detected in WordPress. For private repositories, configure the optional GitHub Personal Access Token under PressHub AI → Settings → Providers.

Reliable Background Scheduling (WP-Cron)

For production sites, high-reliability background research and automated daily briefing harvesting should be bound to a system cron daemon rather than relying on web visitor traffic:

# In wp-config.php:
define( 'DISABLE_WP_CRON', true );

# In crontab (running every minute):
* * * * * wp cron event run --due-now --path=/var/www/html >/dev/null 2>&1

📄 License

This plugin is licensed under the GPL-2.0-or-later license. See the license declaration in presshub-ai-editor/presshub-ai-editor.php for complete terms.

Name		Name	Last commit message	Last commit date
.github/workflows		.github/workflows
.superpowers/sdd		.superpowers/sdd
dev-env		dev-env
docs		docs
presshub-ai-editor		presshub-ai-editor
presshub-workflow		presshub-workflow
.editorconfig		.editorconfig
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
agents.md		agents.md
architecture.md		architecture.md
mcp_config.json		mcp_config.json

Navigation Menu

Repository navigation

About

Repository files navigation

PressHub AI Co-Pilot

🌟 Overview

🚀 Core Subsystems & Features

1. 📰 Automated Daily Briefing Pipeline (Morning Edition & Noon News Update)

2. 🎙️ Podcast Producer, Script Editor & Audio Regeneration

3. ✍️ In-Editor Editorial Co-Pilot

4. 🖼️ AI Featured Image Generation & Studio Engine

5. 🎛️ Instruction Presets & Prompt Studio

🏛️ Architectural Principles

⚙️ Settings-First Principle

🛡️ Zero-Silent-Fallback Principle

🧹 Clean-Slate Architecture

🪵 Enterprise Observability & 4-Layer Logging

CLI Observability Cheat-Sheet

⚙️ Settings Reference Guide

💻 Local Development & Testing

Environment Layout

1. Starting the Dev Server

2. Running PHP Unit Tests

3. Running TypeScript & MCP Workflow Tests

4. Running WP-CLI Commands

📦 Requirements & Installation

Requirements

Installation Steps

Automatic Updates

Reliable Background Scheduling (WP-Cron)

📄 License

About

Releases

Packages

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

PressHub AI Co-Pilot

🌟 Overview

🚀 Core Subsystems & Features

1. 📰 Automated Daily Briefing Pipeline (Morning Edition & Noon News Update)

2. 🎙️ Podcast Producer, Script Editor & Audio Regeneration

3. ✍️ In-Editor Editorial Co-Pilot

4. 🖼️ AI Featured Image Generation & Studio Engine

5. 🎛️ Instruction Presets & Prompt Studio

🏛️ Architectural Principles

⚙️ Settings-First Principle

🛡️ Zero-Silent-Fallback Principle

🧹 Clean-Slate Architecture

🪵 Enterprise Observability & 4-Layer Logging

CLI Observability Cheat-Sheet

⚙️ Settings Reference Guide

💻 Local Development & Testing

Environment Layout

1. Starting the Dev Server

2. Running PHP Unit Tests

3. Running TypeScript & MCP Workflow Tests

4. Running WP-CLI Commands

📦 Requirements & Installation

Requirements

Installation Steps

Automatic Updates

Reliable Background Scheduling (WP-Cron)

📄 License

About

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages