Everlky-yyj/content-director

What does content-director do?

Transform your AI Agent into a short-video lead generation director: per-client persistent state machine + scoring gate + blind prediction + Aha viral compounding, zero-dependency Python (15 commands). Open-source agent skill: persistent per-client loop for short-video lead generation — positioning patterns, standard mining, blind prediction, Aha compounding.

Overview

This skill empowers your AI agent to function as a full-fledged director for short-video lead generation campaigns. It maintains a persistent state machine for each client, allowing the agent to track progress through distinct phases: positioning pattern analysis, content standard mining, blind prediction of trending topics, and Aha moment compounding for viral growth. A scoring gate evaluates content ideas based on predefined criteria, ensuring only high-potential concepts are selected for production. The skill is designed to be self-contained with zero external dependencies, making it easy to deploy and maintain.

Key Features

• Per-client persistent loop: Each client has its own state machine that persists across sessions, enabling continuous optimization and follow-up.
• Scoring gate: Filters content ideas using a scoring system that assesses relevance, engagement potential, and alignment with brand positioning.
• Blind prediction: Uses historical data and trends to predict future viral content topics without bias, enhancing content strategy.
• Aha compounding: Leverages successful Aha moment patterns to create a compounding effect, where past wins inform future content creation.
• Zero-dependency Python: Written entirely in Python with no external libraries required, ensuring portability and ease of integration.

How to Use

1. Installation: Clone the repository Everlky-yyj/content-director and include the Python scripts in your agent's environment.
2. Configuration: Define client profiles with their niche, target audience, and initial positioning patterns.
3. Run the loop: Execute the main command to start the per-client state machine. The agent will automatically move through stages, generate content ideas, score them, and output high-potential concepts.
4. Monitor and optimize: Review the scoring results and adjust parameters (like scoring weights or prediction models) to improve performance over time.

Commands

The skill exposes 15 commands for managing the entire pipeline. These include commands for creating/updating client states, running the scoring gate, triggering blind prediction analysis, and generating Aha compounding sequences. Each command is a simple Python function that can be called directly by your AI agent.

Requirements

• Python 3.x (no external packages)
• Basic knowledge of AI agent frameworks (e.g., Claude Code, Cursor, Codex)

Example Workflow

1. Initialize a new client: content_director init --client-id 'client123' --niche 'fitness tips'
2. Run positioning analysis: content_director position --client-id 'client123'
3. Generate and score ideas: content_director generate --client-id 'client123' --count 10
4. Blind prediction: content_director predict --client-id 'client123'
5. Compile Aha moments: content_director aha --client-id 'client123'

Contributing

Contributions are welcome! Please open issues or pull requests on the GitHub repository.

License

This project is licensed under the MIT License - see the LICENSE file for details.