@dohzya/tools

A collection of CLI tools and AI agent skills for markdown manipulation and productivity.

Quick Links: Claude Code Setup · Codex Setup · CLI Tools Setup · Library API

What's in this repo?

This repository provides three ways to use the tools:

1. AI Agent Skills - AI-powered workflows in Claude Code and Codex
2. CLI Tools - Standalone command-line tools (md, wl, recap, dz-review)
3. TypeScript Library - Importable package from JSR

Installation

Choose your preferred method:

Method	Command	Best For
Claude Code	See CLAUDE_SETUP.md	AI-assisted workflows
Codex	See CODEX_SETUP.md	AI-assisted workflows
Homebrew	brew tap dohzya/tools && brew install md wl recap dz-review	macOS/Linux users
mise	mise plugins install mise-tools https://github.com/dohzya/mise-tools.git && mise use -g mise-tools@X.Y.Z	Project-based installs
Deno	deno install -g jsr:@dohzya/tools/markdown-surgeon/cli	Deno users
Library	deno add @dohzya/tools	TypeScript projects

See CLI_SETUP.md for detailed CLI installation options.

Available Tools & Skills

Name	Description	Available As
markdown-surgeon	Surgically edit Markdown files by section without loading entire content	Skill · CLI (md) · Library
worklog	Track work progress with append-only logs and on-demand checkpoints	Skill · CLI (wl) · Library
recap	Build compact project context snapshots	CLI (recap) · Library
dz-review	Inspect and maintain Markdown review conversations and annotations	Skill · CLI · Library
obsidian-journal	Create journal entries in Obsidian for later reference	Skill only
rex-session	Generate structured REX (Post-Mortem) from technical conversations	Skill only
docs-maintainer	Maintain and review source-verified project documentation	Skill only

markdown-surgeon

Manipulate large Markdown files efficiently:

# CLI
md outline document.md
md read document.md "Section Title"
md write document.md "Section" "Content"

// Library
import { parseDocument } from "@dohzya/tools/markdown-surgeon";
const doc = parseDocument(markdown);

Use cases:

• Editing large .md files (READMEs, documentation)
• Updating specific sections without full file rewrites
• Using Markdown as a lightweight database

Commands: outline, read, write, append, empty, remove, search, concat, meta, create

worklog

Track your work progress with structured logging:

# CLI
wl init
wl add "Implemented feature X"
wl checkpoint "v1.0 feature complete"
wl logs

Use cases:

• Logging work progress during development
• Creating checkpoints at milestones
• Generating summaries for reports or PRs

Commands: init, add, trace, logs, checkpoint, done, list, summary

Activates automatically in supported agents when .worklog/ exists or user says "track this".

obsidian-journal

AI agent skill for creating timestamped journal entries in Obsidian.

Use when: storing, saving, or recording information for later reference.

rex-session

AI agent skill for generating structured REX (Retour d'EXpérience / Post-Mortem) documents from technical conversations.

Use when: documenting incidents, project retrospectives, or technical decisions.

docs-maintainer

AI agent skill for maintaining and reviewing project documentation under docs/.

Use when: creating, updating, organizing, or reviewing durable project documentation.

Library Usage

The core functionality is available as a JSR package:

import { parseDocument } from "@dohzya/tools/markdown-surgeon";

const markdown = `# Title\n## Section 1\nContent`;
const doc = parseDocument(markdown);
console.log(doc.sections);

See packages/tools/README.md for complete API documentation.

Development

First-time setup:

bash setup.sh  # Installs mise and Deno

Commands:

task test      # Run tests
task fmt       # Format code
task validate  # Run all checks (fmt + check + lint + test)

Testing local changes:

⚠️ Installed CLI tools (wl, md) use published JSR versions. To test local code:

deno -A packages/tools/worklog/cli.ts <command>
deno -A packages/tools/markdown-surgeon/cli.ts <command>

CLI test environment notes:

• Unset NO_COLOR when running tests that assert ANSI-colored output.
• Unset WORKLOG_TASK_ID when running worklog tests so agent context does not change CLI defaults.

See AGENTS.md for AI agent guidelines when contributing.

Documentation

• CLAUDE_SETUP.md - Claude Code plugin installation
• CODEX_SETUP.md - Codex plugin installation
• CLI_SETUP.md - CLI tools installation (Homebrew, mise, Deno)
• HOMEBREW_SETUP.md - Detailed Homebrew usage
• MISE_SETUP.md - Detailed mise configuration
• RELEASE.md - Release process for maintainers
• packages/tools/README.md - TypeScript library API

License

MIT