Turn Flat Designs into Editable PSD and PPTX with BGGG Skills

You have a finished design as a flat PNG, and now someone needs the layered PSD or an editable deck. bggg-skills, a 580-star open-source Codex skills collection from BGGG, reconstructs both — and ships more than a dozen further skills for TikTok research, scraping, and Amazon keyword work.

Why This Skill Matters

Every skill is a standalone directory you copy or symlink into ~/.codex/skills/. The README catalogs 16: creator skills (bggg-creator-image2psd turns one or more images into layered, editable PSD with color separation and white-to-transparent conversion; bggg-creator-image2ppt turns images, screenshots, HTML, or SVG drafts into editable PPTX with text-box and native-shape reconstruction), a skill evolver (bggg-skill-taotie incrementally upgrades a target skill by absorbing strengths from others), data collectors for Amazon reviews, Reddit, and X, a six-skill TikTok suite (search, download, video-to-readable-structure, 9:16 cutting, CapCut drafts, and a Feishu-to-TikTok video workflow), Sif keyword tools for Amazon sellers, and X research via Xquik — plus web-access, a third-party MIT skill used optionally in the Sif workflow. One drift note from checking the repo on September 15, 2026: the repository carries 17 skill directories — ngs-amazon-image-studio is present but absent from the README's list as of that date.

For new contributions the README recommends a bilingual layout: each skill ships a SKILL.md for Codex to read, README.md (Chinese) and README_EN.md for humans, with scripts/, references/, assets/, evals/, and a projects/ output directory alongside.

Installation

The documented path is clone, copy, and install per-skill dependencies:

git clone https://github.com/binggandata/bggg-skills.git
cd bggg-skills
mkdir -p ~/.codex/skills
cp -R bggg-creator-image2psd ~/.codex/skills/
cp -R bggg-creator-image2ppt ~/.codex/skills/

For development, symlinks work too (ln -s "$PWD/bggg-creator-image2psd" ~/.codex/skills/bggg-creator-image2psd). Where a skill ships scripts/requirements.txt, install it — the README's examples for the two creator skills:

python3 -m pip install -r ~/.codex/skills/bggg-creator-image2psd/scripts/requirements.txt
python3 -m pip install -r ~/.codex/skills/bggg-creator-image2ppt/scripts/requirements.txt

Real Workflow: Rebuild a Design into an Editable Deck

1. Install bggg-creator-image2ppt as above, plus its requirements.
2. Point Codex at your design source — an image, a screenshot, an HTML mock, or an SVG draft.
3. The skill reconstructs the design as an editable PPTX: components are rebuilt with Codex/imagegen assistance, text boxes are restored as text, and shapes come back as native shapes rather than baked images.
4. Need the print-side artifact instead? bggg-creator-image2psd takes one or more images and splits them into layered PSD — per-color separation, white backgrounds converted to transparency, full-canvas PNG layer export — with PSD writing done in pure Python.

Real Workflow: First Run of the Sif Keyword Scout

1. Copy the Sif skills and the example config (the README's install block includes it): cp -R sif-keyword-scout sif-keyword-tracker web-access ~/.codex/skills/ and cp .sif-config.example.json ~/.codex/skills/.
2. Install the Sif dependencies: python3 -m pip install pandas openpyxl matplotlib python-docx numpy.
3. On first run of sif-keyword-scout, the agent copies .sif-config.example.json to a local .sif-config.json and asks where reports should go. The skill processes Sif's three keyword tables, rates them in tiers, analyzes competitor weaknesses and ad gaps, and outputs a PD focus-keyword list plus a Word report.
4. Once an ASIN has history, sif-keyword-tracker compares two PD keyword snapshots inside a 1–7 day window and outputs a report of keyword updates and ad-placement recommendations.

Tips

• projects/ inside each skill is runtime output — the repo keeps only .gitkeep, so your generated PSDs, zips, and TikTok research packages never get committed.
• web-access is a third-party MIT skill from eze-is/web-access, optionally used in the Sif workflow for browser CDP auto-export.
• Generated artifacts stay local by design: the TikTok skills also ignore downloaded videos, screenshots, CSV/JSON research bundles, subtitles, transcriptions, and CapCut drafts in git.
• The per-skill layout (SKILL.md for Codex, bilingual READMEs, evals/) is the repo's documented convention for contributing new skills.

When Not to Use This

The documented install paths target Codex (~/.codex/skills/); the repo documents no Claude Code or other-runtime layout, so you would be adapting paths yourself. The heavier skills assume real local dependencies — Python packages, and for bggg-tiktok-search a logged-in local Chrome. And the creator skills reconstruct designs; they do not replace a designer. If you need pixel-perfect source files, get the originals instead of rebuilding from a flat export.

See the leaderboard for more skills.