with one click
next-todo
Read todo.md and recommend the next task to work on
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Menu
Read todo.md and recommend the next task to work on
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Audit Claude-authored commits across all repositories Claude has worked in against the code-style guide, reporting violations plus guide defects (missing, dead, or ambiguous rules) the violations reveal. Run on demand from the dotfiles (marketplace) repo.
riseshia's coding conventions. Use it always unless a project-specific convention (CLAUDE.md, style guides) exists.
Quality-check the code-style guide for contradictions, duplications, and ambiguous rules. Run after editing the code-style skill files.
Explain a topic as a single, readable, self-contained HTML file that the user opens in a browser. Use when the user runs /explain-with-html, or asks to have something explained "as HTML", "in a browser", or with a visual/diagram-heavy write-up. The topic is optional โ infer it from the recent conversation and confirm before generating. Use mermaid.js for diagrams when they help.
Run a development task as an event-driven state machine (plan โ plan-approve โ implement โ validate โ minor-fix/recheck โ open-pr โ followup โ retrospect โ done). A thin orchestrator fires named transitions via a CLI that enforces the plan-approve human gate and bounded rework/continue loops, reads a per-state prompt for each node, delegates heavy work-states to fresh workers (Agent tool or claude -p) over a file-based handoff, and ships via a Draft PR. Use when the user wants to run a non-trivial task end-to-end with minimal supervision.
High-precision code review inspired by Alibaba Open Code Review. Script-driven pre-processing, per-file isolated review, three-layer false-positive reduction.
| name | next-todo |
| description | Read todo.md and recommend the next task to work on |
| user_invocable | true |
| arguments | priority_filter |
cat todo.md 2>/dev/null || echo "todo.md not found"Based on the todo list above, recommend the next actionable task.
For each task category, evaluate:
If the user provided arguments (e.g., /next-todo P5), filter to that priority level only.
Present your recommendation in Korean:
## ์ถ์ฒ ์์
**์ ๋ชฉ:** [task title]
**์ฐ์ ์์:** [P1-P5]
**์์ ๋ณต์ก๋:** [๋ฎ์/์ค๊ฐ/๋์]
### ์ด์
[Why this task should be done next - 2-3 sentences]
### ์์ ํฌ์ธํธ
[Specific files or steps to begin with]
---
### ๋ค๋ฅธ ํ๋ณด
1. [Alternative task 1] - [brief reason]
2. [Alternative task 2] - [brief reason]