Agent skill

checkpoint-manager

Skill for tracking and managing task progress. Checks pending tasks at conversation start, proposes task creation for new requests, suggests completion reports when work is done. Integrates with checkpoint system.

View SKILL.md on GitHub Repository
Stars 1
Forks 0

Install this agent skill to your Project

npx add-skill https://github.com/dobachi/AI_Instruction_Kits/tree/main/templates/claude-skills/en/checkpoint-manager

SKILL.md

Checkpoint Manager Skill

A skill for consistently tracking and reporting task progress.

Auto-Suggestion Triggers

Situation Suggestion
Conversation start "Shall I check for pending tasks?"
New task request "Shall I start a checkpoint?"
Before loading instructions "Shall I record instruction usage?"
Work milestone "Shall I report progress?"
Work completion "Shall I complete the task?"

Workflow Overview

┌─────────────┐    ┌─────────────┐    ┌─────────────┐    ┌─────────────┐
│  pending    │    │  start      │    │  progress   │    │  complete   │
│  check      │ → │  begin      │ → │  report     │ → │  finish     │
└─────────────┘    └─────────────┘    └─────────────┘    └─────────────┘
                          ↓
                   ┌─────────────┐
                   │ instruction │
                   │ management  │
                   └─────────────┘

See workflow.md for details.

Command Reference

Task Management

bash
# Check pending tasks
scripts/checkpoint.sh pending

# Start a new task (task ID auto-generated)
scripts/checkpoint.sh start "<task-name>" <steps>

# Report progress (only during instruction use)
scripts/checkpoint.sh progress <task-id> <current> <total> "<status>" "<next>"

# Complete task (all instructions must be completed)
scripts/checkpoint.sh complete <task-id> "<result>"

# Report error
scripts/checkpoint.sh error <task-id> "<message>"

Instruction Management

bash
# Start using an instruction
scripts/checkpoint.sh instruction-start "<path>" "<purpose>" [task-id]

# Complete instruction use
scripts/checkpoint.sh instruction-complete "<path>" "<result>" [task-id]

Status Check

bash
# Show task history
scripts/checkpoint.sh summary <task-id>

Usage Scenarios

Scenario 1: Conversation Start

When user starts a new conversation, suggest checking for pending tasks:

AI: Shall I check for pending tasks?

# Execute
scripts/checkpoint.sh pending

Scenario 2: New Task Request

When user requests something like "implement X":

AI: Shall I start a checkpoint?
    Task name: [user's request]
    Estimated steps: [based on complexity]

# Execute
scripts/checkpoint.sh start "Feature implementation" 5
# → Task ID: TASK-123456-abc123

Scenario 3: Instruction Use

Before loading an instruction:

AI: Shall I record instruction usage?

# Execute
scripts/checkpoint.sh instruction-start "instructions/en/coding/web_api.md" "REST API development" TASK-123456-abc123

Scenario 4: Work Milestone

When significant progress is made:

AI: Shall I report progress?
    Current: 2/5 steps
    Status: Design complete
    Next: Start implementation

# Execute
scripts/checkpoint.sh progress TASK-123456-abc123 2 5 "Design complete" "Start implementation"

Scenario 5: Work Completion

When task is complete:

AI: Shall I complete the task?
    Result: [summary of work]

# Execute instruction completion first
scripts/checkpoint.sh instruction-complete "instructions/en/coding/web_api.md" "3 endpoints implemented" TASK-123456-abc123

# Then complete task
scripts/checkpoint.sh complete TASK-123456-abc123 "REST API 3 endpoints implemented"

Workflow Constraints

Constraint Description
Progress report Only allowed during instruction use
Task completion All instructions must be completed first
Task ID omission Warning displayed for instruction commands

Decision Criteria

When to Suggest Task Start

  • User requests specific work
  • Work involves multiple steps
  • Keywords like "implement", "create", "fix"

When to Suggest Progress Report

  • Significant code changes made
  • Tests passed/failed
  • Clear work milestone reached

When to Suggest Task Completion

  • User says "done", "finished", "complete"
  • All requirements met
  • Tests passing

Notes

  • Task IDs are auto-generated (e.g., TASK-123456-abc123)
  • Keep status and actions brief and clear
  • Use the same task ID throughout a task
  • Run scripts/checkpoint.sh from project root

Recommended Agent Skills

Expand your agent's capabilities with these related and highly-rated skills.

dobachi/AI_Instruction_Kits

auto-build

プロジェクトタイプを自動検出してビルドを実行するスキル。コード変更後にビルドを提案、エラー発生時は修正案を提示。Node.js、Rust、Python、Go、Makefileプロジェクトに対応。

1 0
Explore
dobachi/AI_Instruction_Kits

checkpoint-manager

タスクの進捗を追跡・管理するスキル。会話開始時に未完了タスクを確認、新規タスク依頼時にタスク開始を提案、作業完了時に完了報告を提案。チェックポイントシステムと連携。

1 0
Explore
dobachi/AI_Instruction_Kits

commit-safe

安全なコミットを支援するスキル。変更内容を確認してから選択的にコミット。大きな変更時にファイル指定コミットを提案、git add -Aの使用を防止。

1 0
Explore
dobachi/AI_Instruction_Kits

worktree-manager

Git worktreeの作成・管理を行うスキル。複雑なタスクや複数ファイル変更時にworktree作成を提案、タスク完了時にマージ・クリーンアップを案内。checkpoint-managerと連携。

1 0
Explore
dobachi/AI_Instruction_Kits

auto-build

Skill for auto-detecting project type and executing builds. Suggests builds after code changes, proposes fixes for build errors. Supports Node.js, Rust, Python, Go, and Makefile projects.

1 0
Explore
dobachi/AI_Instruction_Kits

commit-safe

Skill for safe commits. Reviews changes before selective commits. Proposes file-specific commits for large changes, prevents use of git add -A.

1 0
Explore

Didn't find tool you were looking for?

Be as detailed as possible for better results