binggandata/bggg-skills

Open-source Codex skills from BGGG

What it solves

This repository bundles a set of reusable skills that extend Codex (an AI coding agent) to handle specific creative, data, and video tasks. Instead of writing one-off prompts or scripts, users can copy a skill folder into Codex's skills directory and get a repeatable, well-structured workflow. The skills address common gaps like converting images or HTML into editable PSD/PPTX files, scraping public data from Amazon/Reddit/X/TikTok, turning video content into readable metadata and transcripts, cutting videos for TikTok, analyzing Amazon keyword data, and performing X research through API keys.

How it works

Each skill lives in its own folder with a standard layout: a SKILL.md file that Codex reads to understand when and how to trigger the skill, a README.md (Chinese) and README_EN.md (English) for human instructions, scripts/ for deterministic code, references/ for optional reading material, assets/ for static resources, evals/ for test cases, and a projects/ directory that holds runtime outputs (which are gitignored). To install, you clone the repo and either copy or symlink the skill folders into ~/.codex/skills/. Some skills require additional Python package dependencies listed in their scripts/requirements.txt files. The Sif keyword workflow uses a config file (.sif-config.example.json) that gets copied to a local .sif-config.json on first run. The collection currently includes 16 skills, ranging from image-to-PSD/PPT converters, to TikTok video processing tools, to data collectors for Amazon/Reddit/X, plus an 'evolution' skill that improves other skills by analyzing their strengths.

Who it's for

People who use Codex as their AI assistant and want ready-made, battle-tested workflows for design file creation, social media content production, data collection, and market research. It's beginner-friendly for Codex users because installation is just copying a folder, but it also helps advanced users who want to inspect or contribute skills.

Highlights

  • Offers a broad set of 16 distinct skills covering image/design conversion, video processing, data scraping, and keyword analysis.
  • Each skill is self-contained, with clear documentation (Chinese and English) and a standard folder structure that makes them easy to copy, symlink, or extend.
  • Includes deterministic scripts and reference material to ensure consistent behaviour.
  • Several skills are specifically tailored for TikTok workflows (search, download, read, cut, and CapCut drafting) and for Amazon keyword analysis (Sif scout and tracker).
  • A dedicated 'evolution' skill (taotie) helps other skills improve by comparing and absorbing best practices.
  • Runtime outputs are kept out of the repository, reducing clutter and privacy risk.

Related

  • Project
    XYZ-AI-Lab/axrl

    An agentic RL post-training framework that coordinates high-throughput rollouts via SGLang and large-scale training via Megatron for multi-turn agent workflows.

  • Project
    zhudotexe/kani

    Kani is a lightweight, flexible framework for building chat-based language model applications that simplifies function calling and context management without enforcing opinionated prompt structures.

  • Project
    wbh604/fund-guy-skill

    A behavioral audit engine for fund managers that analyzes years of trading data to determine if returns were driven by skill or market luck, producing detailed interactive reports.

  • Project
    berabuddies/Semia

    A security auditing tool for AI agent skills that analyzes markdown-based skills to identify potential security risks and provide evidence-backed reports before execution.