blank-agentic-cli 0.1.8

CLI scaffolder for agentic AI research projects

blank

blank is a CLI scaffolder for research projects driven by agentic AI workflows.

Install

Recommended:

curl -fsSL https://raw.githubusercontent.com/shusuke-ioku/blank/main/scripts/install_blank.sh | bash

Alternative: manually install:

1. Install pipx (if missing):

python3 -m pip install --user pipx
python3 -m pipx ensurepath

2. Reload your shell:

exec $SHELL -l

3. Install blank from PyPI:

pipx install blank-agentic-cli

Verify:

blank --help

Troubleshooting

If blank --help fails with command not found, run:

export PATH="$HOME/.local/bin:$PATH"
hash -r
blank --help

Make it persistent:

# zsh
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zprofile

# bash
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bash_profile
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc

Debug checks:

pipx list
echo "$PATH" | tr ':' '\n' | rg "$HOME/.local/bin"
which blank
ls -l "$HOME/.local/bin/blank"

Usage

blank init
blank init my-project
blank init my-project --project-name "My Project"
blank init my-project --dry-run
blank init my-project --force
blank init my-project --no-agents

By default, paper/paper.typ imports TeXst:

#import "@preview/texst:0.1.0": paper, nneq, caption-note, table-note, theorem, proof, prop, lem, rem, ass, cmain, csub, caption-with-note

TeXst repository:

• https://github.com/shusuke-ioku/texst

paper/aesthetics.typ is intentionally generated as a blank file so you can add custom style settings.

Generated scaffold

• analysis/scripts/
• analysis/data/
• analysis/output/
• paper/
• idea/
• .codex/project.md with default agent rules
• .codex/config.toml with Zotero + PDF-reader MCP server setup
• .codex/skills/ starter skill set and install scripts
• .claude/ defaults

MCP setup (Zotero + PDF reader)

Generated projects include:

• zotero MCP server via zotero-mcp
• pdf_reader MCP server via uvx markitdown-mcp

Quick local checks:

zotero-mcp --help
uvx markitdown-mcp --help

Development

python3 -m venv .venv
source .venv/bin/activate
pip install -e . pytest
pytest

Release (PyPI)

1. Create PyPI project blank-agentic-cli and enable Trusted Publishing for this GitHub repo.
2. (Optional) Run GitHub Action publish manually with testpypi to verify packaging.
3. Tag a release:

git tag v0.1.0
git push origin v0.1.0

4. The publish workflow builds and uploads to PyPI.