text-poster-studio
v1.0.1
Published
Codex skill and CLI for compiling source material into editable text-poster and social-card projects.
Maintainers
Readme
Text Poster Studio
An agent skill for turning source material into schema-valid, fully editable social-card and text-poster projects.
The agent handles research boundaries, content strategy, and the first structured draft. Deterministic scripts compile and validate the result before it is handed to a compatible visual editor.

What it does
- Creates card decks from text, notes, documents, or URLs.
- Adapts an existing deck to another audience, channel, or visual direction.
- Separates facts, interpretation, examples, and recommendations.
- Supports progressive domain context with explicit provenance notes.
- Emits editable project JSON instead of flattened artwork.
- Validates page counts, content roles, canvas presets, themes, and design tuning.
Install from npm
npx text-poster-studio installThis copies the skill into ~/.codex/skills/text-poster-studio. Start a new Codex session so the skill catalog reloads.
Install from GitHub
git clone https://github.com/aa2246740/text-poster-studio.git
cp -R text-poster-studio/skills/text-poster-studio ~/.codex/skills/Use
$text-poster-studio Turn this article into a six-page LinkedIn carousel
for product leaders. Preserve factual uncertainty and return an editable
project JSON.For Chinese content:
$text-poster-studio 把这份文章整理成 7 张小红书观点卡,
面向 AI 产品经理,允许补充背景知识并标注来源。The output flow is:
source → agent content plan → deterministic compiler → schema validation
→ editable project JSON → visual editorThe current schema is compatible with Viewpoint Cards Studio schema v2. The compiler itself uses only Node.js built-in modules.
Example: Seeing like an Agent
The included example starts with a long-form summary of an Agent tooling talk and turns it into a nine-card Chinese deck for AI product managers and developers.
$text-poster-studio 把这份演讲整理成小红书观点卡,
面向 AI 产品经理与 Agent 开发者,不补写未经来源支持的事实。Inspect or reuse the full example:
The screenshots above are real 1242×1656 PNG exports from the editable project.
CLI
# Install the Codex skill
npx text-poster-studio install
# Inspect supported structures, channels, themes, and layouts
npx text-poster-studio options
# Compile an Agent-authored content plan
npx text-poster-studio compile \
--input examples/seeing-like-an-agent/content-plan.json \
--output deck.text-poster.json
# Validate an editable project
npx text-poster-studio validate deck.text-poster.jsonRepository layout
bin/
└── text-poster-studio.mjs
docs/images/
examples/seeing-like-an-agent/
skills/text-poster-studio/
├── SKILL.md
├── agents/openai.yaml
├── references/
│ ├── content-plan.md
│ └── project-contract.md
└── scripts/
├── compile-project.mjs
├── project-contract.mjs
└── validate-project.mjsValidate
python3 ~/.codex/skills/.system/skill-creator/scripts/quick_validate.py \
skills/text-poster-studio
node skills/text-poster-studio/scripts/compile-project.mjs --list-options
npm pack --dry-runLicense
MIT
