Skip to main content
skill CX Codex

next · mag123c — Claude Code skill: project status and next task

A session-start skill: reads your latest planning file, checks the git log, and tells you what to work on next.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add mag123c/toktrack --skill next -a claude-code
Open on GitHub

What next does

Run next at the start of a session when you need to pick up where the project left off. It opens only the newest planning file, matched by the YYYYMMDD- date prefix, and reads the checkbox states. Then it runs git log --oneline -5 and git status --short.

From there it counts finished tasks against the total, works out the current phase, and picks the next task by priority. The output is a short block of 5–10 lines: a phase table with status and progress, a summary of the next task, and a pointer to /clarify. With no planning files around, the skill falls back to the git log and the state of the code. It expects .claude/ai-context/architecture.md to exist; without that file the architecture context stays empty.

On Claude, install with npx skills add mag123c/toktrack --skill next -a claude-code — the skill lands in ~/.claude/skills/next/SKILL.md. Codex, Cursor, GitHub Copilot, Gemini CLI and OpenCode are supported too. Copilot uses a different command: gh skill install mag123c/toktrack next.

What it helps with

  • figure out where a project stopped
  • count finished tasks in a planning file
  • pick the next task by priority
  • compare repo state against the plan
  • start a session without rereading notes by hand

Example prompts for the agent

  • Check where we left off and suggest the next task
  • Read the latest planning file and show progress by phase
  • What should I work on next?

How to install next

1Skills CLI (Vercel)

npx skills add mag123c/toktrack --skill next -a claude-code

2Manually: put the files in place

~/.claude/skills/next/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/next/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about next

How do I install next in Claude Code?

Run npx skills add mag123c/toktrack --skill next -a claude-code. The skill is placed at ~/.claude/skills/next/SKILL.md.

What does the next skill do?

It reads the newest planning file and the git log, counts task progress, and names the next task along with a suggestion to run /clarify.

What files does next need?

A planning file with a date prefix, plus .claude/ai-context/architecture.md. When no planning files exist, the skill works from the git log and the code itself.

Source text

SKILL.md · MIT
---
name: next
description: Session start - check progress, suggest next task
required_context:
  - .claude/ai-context/architecture.md
---

# Next

## Flow
```
Read Planning → Git Log → Analyze → Present → Suggest /clarify
```

## Execution

1. **Read Planning**
   ```bash
   # Read ONLY the latest planning file (by date prefix YYYYMMDD-)
   # e.g., 20260205-improvements.md > 20260128-cli-parsers.md
   # Check checkbox status: [ ] incomplete, [x] complete
   ```

2. **Git Log**
   ```bash
   git log --oneline -5
   git status --short
   ```

3. **Analyze**
   - Identify current phase
   - Count completed/total tasks
   - Identify next priority task

4. **Present** (table format)
   | Phase | Status | Progress |
   |-------|--------|----------|
   | Phase 0 | ✅ | 5/5 |
   | Phase 1 | 🔄 | 3/4 |

5. **Suggest**
   - Summarize next task
   - Suggest running `/clarify`

## Output Format
```markdown
## Current Status
- Phase: {current_phase}
- Progress: {completed}/{total} tasks

## Next Task
**{task_id}: {task_name}**
{brief_description}

## Action
Run `/clarify` to start: {task_summary}
```

## Rules
- **Read only the latest dated planning file** (highest YYYYMMDD- prefix)
- If no planning files → infer from git log + code state
- Keep output concise (5-10 lines)
- Always suggest /clarify connection

© mag123c · MIT · mag123c/toktrack

Similar skills

All skills for Agent setup: memory, context, orchestration →

next by other authors comes first.

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor