Eisenhower Task Manager

Task management based on Eisenhower Matrix + P0-P2 priority with Customer Project Management. Four quadrants for execution, separate Customer Project List fo...

Security Scan

The skill generally matches a local markdown-based task manager, but the package metadata/instructions disagree with the included code (it requires Node/npm and runs a local dashboard that reads and modifies files), which is an incoherence you should review before installing.

Purpose & Capability

The skill claims to be instruction-only with no required binaries/config, but the bundle contains a Node.js dashboard (server.js, parser.js, start.sh, check-and-install.sh, package.json). Running the dashboard requires Node/npm and will read/write files under a tasks/ directory — reasonable for a task manager but inconsistent with the declared 'no required binaries' and 'instruction-only' metadata.

Instruction Scope

SKILL.md instructs the agent to read, modify, renumber, delete, and archive markdown task files (tasks.md, customer-projects.md, maybe.md, delegation.md, archived.md) after every operation. Those are destructive file operations (delete/move/renumber). It also directs offering to 'launch dashboard and open browser' which implies starting a local server and opening a browser. The file modifications and auto-launch behavior are within the skill's purpose but are high-impact and must be explicitly consented to.

Install Mechanism

There is no formal install spec in the registry metadata, but the package includes start/check-and-install scripts that run 'npm install' and depend on common npm packages (express, ws, chokidar). Installing pulls from the public npm registry (package-lock.json shows registry URLs). This is a normal flow for a Node app but the registry entry should have declared this dependency and required binaries (node/npm).

Credentials

The skill does not request environment variables, credentials, or external endpoints. Code reads local markdown files from a tasks/ directory and uses local npm packages — this is proportionate to its stated purpose.

Persistence & Privilege

always:false (normal) and no declared credential persistence. However, the dashboard starts a local HTTP/WebSocket server that will serve the parsed task data and watch files for changes (chokidar). You should confirm the server binds only to localhost and does not expose files to the network unintentionally.

What to consider before installing

Things to check before installing or running this skill: - Review server.js and start.sh: confirm the server listens on localhost (127.0.0.1) rather than 0.0.0.0, and verify there are no unexpected external network calls or endpoints. - Confirm you have Node.js/npm and are comfortable the start script will run 'npm install' in the skill directory (it will download packages from the public npm registry). - Inspect the code that reads/writes files (parser.js and whatever performs renumber/delete/archive operations) to ensure it only touches the intended tasks/*.md files and does not access other paths. - Be aware the skill will modify your markdown files (delete/move/renumber) as part of 'Complete Task' and 'Auto-fix numbering' flows — back up your tasks/ directory before first run. - If you cannot review the server code, run it in a sandboxed environment or container to limit filesystem and network exposure. - The registry metadata should declare required binaries (node/npm) and an install spec; the absence of these is an inconsistency to flag with the skill author.

Task Manager - Four-Level System

Architecture

Layer	File	Purpose
Customer	tasks/customer-projects.md	ALL customer projects (single source of truth)
Execution	tasks/tasks.md	Four quadrants (personal + customer tasks requiring YOUR execution)
Future	tasks/maybe.md	Ideas for later evaluation
Delegation	tasks/delegation.md	Tasks assigned to others
Archive	tasks/archived.md	Completed/deleted history

Core Principles

1. Customer projects NEVER duplicate - exist only in Customer Project List
2. Personal execution only - customer tasks enter quadrants only when YOU must execute
3. Sequential numbering - 1-N per file, no gaps, no duplicates
4. Auto-fix numbering - after EVERY operation, verify and fix sequence

Four Quadrants

Q	Name	Strategy	When
Q1	Important + Urgent	Execute immediately	Deadlines, blockers
Q2	Important + Not Urgent	Plan & schedule	OKRs, strategic work
Q3	Not Important + Urgent	Batch process	Maintenance, delegatable
Q4	Not Important + Not Urgent	Postpone/delete	Exploration, optional

Priority Tiers (within quadrant)

• P0: Today (hard deadline or blocking)
• P1: This week
• P2: This month/quarter

Customer Project Workflow

New Customer Project
    ↓
Add to Customer Project List (record all)
    ↓
Need YOUR execution?
    ├─ YES → Add to appropriate Quadrant, tag with [Customer/ProjectName]
    └─ NO → Keep in Customer Project List only
    ↓
Regular Review → Update status (Active/Blocked/Pending)
    ↓
Completed → Archive entry + complete quadrant task if exists

Task Numbering Rules

Quadrant Tasks (tasks.md)

• Q1: 1 to N
• Q2: (N+1) to M
• Q3: (M+1) to P
• Q4: (P+1) to Q
• Continuous sequence across quadrants, no gaps

Maybe List (maybe.md)

• Sequential: 1, 2, 3...
• ALWAYS append to END - never insert in middle
• Verify: last_number == total_count

Customer Projects (customer-projects.md)

• Per-customer numbering: 1, 2, 3...
• Independent across customers

Delegation List (delegation.md)

• Sequential: 1, 2, 3...
• ALWAYS append to END
• Verify: last_number == total_count
• Format: ### X. Task Name @assignee

Archive (archived.md)

• Sequential: 1, 2, 3...
• ALWAYS append to END
• Group by archive date
• Format: ### X. Task Name [Archived: YYYY-MM-DD]

Numbering Verification (MUST RUN AFTER EVERY OPERATION)

1. Read target file
2. Scan for: duplicates, gaps, wrong transitions
3. Reassign sequential numbers starting from 1
4. Re-read to verify
5. Repeat if issues remain

Numbering Checklist:

• Sequential from 1, no gaps
• No duplicates
• Q2 starts at Q1_count+1, Q3 at Q2_end+1, etc.

Operations

Add Task

See references/task-add.md for complete workflow.

Quick Guide:

1. Determine level: Quadrants / Delegation / Maybe
2. Determine quadrant (if applicable): Q1/Q2/Q3/Q4
3. Assign priority: P0/P1/P2
4. Insert at appropriate position
5. Renumber ALL tasks (sequential from 1)
6. Update statistics
7. Display updated list
8. (Optional) Offer to open Dashboard for visual review

Complete Task

See references/task-complete.md for complete workflow.

Key Points:

• MUST NOT stop at just marking [x] - full flow required
• Delete completed task from source file
• Renumber remaining tasks
• Move to archived.md
• Update statistics
• Display updated list
• (Optional) Offer to open Dashboard to see updated status

Three Scenarios:

1. Main Task List (tasks.md) - Four Quadrant tasks
2. Delegation List (delegation.md) - Delegated tasks
3. Maybe List (maybe.md) - Future tasks

Post-Operation Dashboard Offer

After any task operation (add/complete/update), the skill SHOULD offer to open the dashboard:

✅ Task Added Successfully

**Summary:**
- Task: [Task Name]
- Location: Q2 #15
- Priority: P1

📊 Would you like to view the task dashboard?
→ User approves: Launch dashboard and open browser
→ User declines: Continue with next operation

Implementation: Check if user wants to open dashboard after major operations.

Add Customer Project

Step 1: Add to Customer Project List

### [Customer Name]

#### X. Project Name [Status]
- **Status**: Active / Blocked / Pending / Completed
- **Type**: Implementation / Support / POC / Maintenance
- **Description**: Brief
- **Created**: YYYY-MM-DD
- **Last Review**: YYYY-MM-DD
- **Notes**: Progress, blockers

Step 2: If personal execution needed, also add to Quadrants

Complete Customer Project

1. Mark Completed in Customer Project List
2. Archive the entry in archived.md
3. Complete related quadrant task if exists

Add to Maybe List

1. Read tasks/maybe.md
2. Find highest task number (N)
3. Append new task as #(N+1) at END
4. Update total count and timestamp

Check/Fix Numbering

Trigger phrases: "check task numbering", "fix sequence", "序号问题"

Action: Read file → scan issues → auto-fix → re-verify → repeat if needed

Output Format (MUST OUTPUT AFTER EVERY OPERATION)

✅ [Operation] Completed

**Summary:**
- Operation: [Add/Update/Complete/Move/Delete]
- Target: [Task name/ID]
- Location: [Q1/Q2/Q3/Q4/Customer/Maybe/Delegation/Archive]

**Statistics:**
| Quadrant | Count |
|----------|-------|
| Q1 | X |
| Q2 | X |
| Q3 | X |
| Q4 | X |

See references/output-examples.md for detailed examples.

Healthy Metrics

List	Target	Warning
Customer Projects	5-15	-
Q1	3-5	>5 indicates poor planning
Q2	8-12	Core investment area
Q3	5-10	Automate gradually
Q4	<10	Clean up regularly
Maybe	<20	Review monthly
Delegation	<15	Review weekly

Time Rhythm

• Daily: Check Q1, customer escalations
• Weekly: Review Customer Projects, Q2 planning, Delegation check
• Monthly: Archive completed, review blocked, adjust priorities

Task Tags

• Customer tasks in quadrants: [Customer/ProjectName]
• Blocked tasks: 🚫 Blocked
• Status: P0/P1/P2

Dashboard (Optional)

A real-time web dashboard is available to view all tasks in a friendly format.

Location: dashboard/

Features:

• 📊 Real-time statistics showing task counts by quadrant
• 🔥 Four-quadrant matrix view with color coding
• 🏢 Customer projects overview with filtering
• 👑 Delegation list with assignee and deadline info
• 🌱 Maybe list for future ideas
• 🎨 Dark theme with modern UI design
• 🌐 Bilingual support (English/Chinese)
• ⚡ Real-time updates via WebSocket when markdown files change
• 📱 Responsive layout for different screen sizes
• 💡 Hover to see full task details - Mouse over any task card to see complete information including description, subtasks, tags, and timestamps

Dashboard Launch Workflow

After completing task operations, the skill will offer to open the dashboard:

✅ [Operation] Completed

Would you like to view the task dashboard?
[Open Dashboard] [Skip]

If approved → Auto-start dashboard and open browser
If skipped  → Dashboard remains available for manual launch

Manual Launch (Anytime)

Method 1: Daemon Mode (Recommended for Long-term Use)

Run in background, service continues after terminal closes:

cd dashboard
./start.sh --daemon     # Start in background, no terminal dependency

Features:

• ✅ Service continues after terminal closes
• ✅ Auto PID management, prevents duplicate starts
• ✅ Logs output to dashboard.log
• ✅ Shows access URL and PID after startup

Stop the daemon:

./stop.sh               # Gracefully stop the service

Method 2: Foreground Mode (For Debugging)

Suitable for development and debugging, service stops when terminal closes:

cd dashboard
./start.sh              # Run in foreground, Ctrl+C to stop

Advanced Usage

# Specify port (default 8080)
./start.sh --daemon --port 3000

# Check running status
curl http://localhost:8080/api/health

Architecture

• Markdown files are the single source of truth
• Node.js server parses markdown in real-time
• WebSocket pushes updates to browser when files change
• No database required - pure markdown-driven
• Enhanced error handling: global exception capture, port conflict detection

Reference Materials

File	Purpose
references/task-add.md	Complete workflow for adding tasks - levels, quadrants, priorities, renumbering
references/task-complete.md	Complete workflow for completing tasks - 3 scenarios (Main/Delegation/Maybe), archiving
references/numbering-rules.md	Numbering rules for all 5 task lists
references/output-examples.md	Output format examples
references/maybe-list-workflow.md	Step-by-step Maybe List operations
references/templates.md	Statistics and report templates
references/dashboard-offer.md	Dashboard offer workflow after task operations
dashboard/	Real-time web dashboard (./start.sh --daemon for background mode)

Critical: Always read task-add.md or task-complete.md before performing those operations to ensure full workflow compliance.

---

Version: 8.2 (Added daemon mode, stop.sh, enhanced error handling) Last Updated: 2026-04-21