# Auto Target Tracker

> Automatic goal progress tracker. When target-related images (notes, progress, screenshots, records) are detected in the conversation, VLM is automatically called to identify key information and recorded to the target diary. It is suitable for all goal management scenarios such as learning management, fitness tracking, work progress, habit development, and creative records.

- Skill: `zeroix07/auto-target-tracker` (Agent Skill)
- Install (CLI): `npx skillmds@latest add zeroix07/auto-target-tracker`
- Raw SKILL.md: https://api.skillmd.com/api/skills/zeroix07/auto-target-tracker/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: zeroix07 (https://skillmd.com/u/zeroix07)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/zeroix07/auto-target-tracker

---


# Automatic goal progress tracker

## Trigger conditions

Automatically triggered when the following conditions occur in the conversation:

1. **User sent pictures** (especially study notes, progress screenshots, fitness records, task lists, creative works, etc.).
2. **The user sent the picture during the set target time period** (such as 08:30, 10:00, 20:00).
3. **The user clearly said** "help me remember it", "check the progress", "check in", "update it", etc.

---

## Workflow

### 1. Detect pictures

When an image is detected, check:
- Whether the image file name contains the target keyword (progress, goal, task, workout, note, etc.)
- Whether the image content contains target elements (progress bar, text, code, charts, schedules, etc.)
- Whether it is near the scheduled target reminder time
- Whether the user's most recent conversation context involves the execution of the goal

### 2. Call VLM identification

Use the vlm tool to identify images:

**Universal prompt template**:
```
"Identify the key information in the picture and extract the following content according to the target type:
- Core mission/content
- Completion progress or quantity
- Key data (such as time, weight, word count, etc.)
- Give a brief performance feedback"
```

**Target type specific prompt**:

| Target Type | Prompt |
|---------|--------|
| Study | "Identify study notes, extract knowledge points, and completeness" |
| Fitness | "Identify fitness records and extract exercise type, number of groups, times, and weight" |
| Work | "Identify work progress, extract completed tasks and completion rate" |
| Creation | "Identify creative works, extract creation type, progress, and key elements" |
| Habit | "Identify punch-in records, extract punch-in content and consecutive days" |

### 3. Parse target information

Extracted from the results returned by VLM:
- **Task/Content List**: Specific actions or tasks identified
- **Completion**: Progress estimate based on picture content
- **Key data**: Quantitative indicators such as time, quantity, weight, word count, etc.
- **Cognitive Feedback**: A brief review of the current goal state

### 4. Record to target diary

Call the `edit_daily` tool to record the recognition results into the daily notes of the day


### 5. Feedback to users

Confirm the recognition result to the user:

```
Your goal check-in has been recorded:

📝 Recognition results:
Core content: What you took was today’s English word list, and you memorized a total of 15 new words.
Progress estimate: Today’s vocabulary tasks are all completed, and the progress is better than 80% of the study parties.
Suggestion: The spelling of two words is a bit unclear. Remember to take a closer look when you review tomorrow.

Are the records accurate? Do you want me to save it in your goal diary for today?
```

---

## Record format

### Target Journal Entry Example

```markdown
## 20:00 punch-in record

**Goal Type**: 📚 Learning

**Image**: ![target image](path/to/image.jpg)

**VLM recognition results**:

| Task/Content | Progress/Quantity | Status |
|----------|----------|------|
| English words (Unit 1) | 15 | Completed |
| Math Exercises (Chapter 3) | 80% | In progress |

| **Total** | | **Achieved today 2/3** |

**Key data**:
- Study duration: 2 hours
- Concentration: High

**Remarks**: Automatic recognition, user confirmation is correct

---

## 10:30 Fitness check-in

**Goal Type**: 🏃 Fitness

**Picture**: ![Fitness Record](path/to/gym.jpg)

**VLM recognition results**:

| Type of exercise | Number of sets | Reps | Weight | Condition |
|---------|------|------|------|------|
| Bench Press | 4 | 12 | 60kg | ✅ Complete |
| Squats | 4 | 10 | 80kg | ✅ Complete |
| Pull-ups | 3 | 8 | Body weight | ⚠️ One less set |

| **Total** | | | **Today's standard achieved** |

**Key data**:
- Total weight: 2640kg
- Training duration: 45 minutes

**Remarks**: One less set of pull-ups needs to be completed next time.
```

---

## Integration with target system

### Daily summary

In the summary every night at 22:00, it includes:
- All punch-in records today
- Goal achievement rate analysis
- Comparison with target (if target is set)

### Weekly/Monthly Report

In the weekly report, include:
- Effective execution time this week
- Target coverage
- Number of consecutive check-in days
- Dynamic difficulty adjustment suggestions: If the target is reached continuously, it is recommended to increase the task volume next week

---

## Common usage scenarios

### Scenario 1: Learning to clock in

**User Behavior**: Send photo of handwritten note

**Automatic recognition**:
- Extract knowledge points
- Calculate learning progress
- Record to learning log

**Feedback Example**:
```
📚 Recognized study notes:
- Machine learning supervised learning algorithm (completed)
- Gradient descent optimizer (in progress)
- Regularization to prevent overfitting (not started yet)

Progress: 33% | Estimated 2 hours to complete
```

### Scene 2: Fitness check-in

**User Behavior**: Send fitness record photos

**Automatic recognition**:
- Extract motion type
- Statistics of sets, reps, and weight
- Calculate training volume

**Feedback Example**:
```
🏃 Fitness records identified:
- Bench Press 60kg × 12 × 4 groups ✅
- Squat 80kg × 10 × 4 groups ✅
- Pull-ups Self-weight × 8 × 3 groups ✅

Total training volume: 2640kg | Duration: 45 minutes
```

### Scenario 3: Work progress

**User Behavior**: Send project progress screenshots

**Automatic recognition**:
- Extract completed tasks
- Calculate completion percentage
- Identify remaining tasks

**Feedback Example**:
```
💼 Work progress identified:
- Requirements document (completed)✅
- Prototype design (completed)✅
- Front-end development (in progress) 🔄 80%
- Backend development (not started yet) ⏳

Overall project progress: 67%
```

### Scene 4: Creation check-in

**User Behavior**: Send photos of creative works

**Automatic recognition**:
- Extract creation type
- Identify key elements
- Estimate completion

**Feedback Example**:
```
🎨 Creation record identified:
Type: illustration creation
Elements: characters, background scenes
Completion: 100% line drawing, 60% coloring

Suggestion: I finished the character line drawing today and can start coloring the background tomorrow
```

### Scenario 5: Get used to clocking in

**User Behavior**: Send screenshot of check-in calendar

**Automatic recognition**:
- Extract the number of consecutive check-in days
- Identify today's check-in status
- Calculate punch rate

**Feedback Example**:
```
✅ Habit of punching in has been recognized:
Get up early: 15 consecutive days | Check-in rate 100%
Reading: 8 consecutive days | Check-in rate 73%
Sports: 21 consecutive days | 100% check-in rate

🎉 Exercise has been clocked in for 3 consecutive weeks, keep it up!
```

---

## Scope

This skill ONLY:
- Identify target-related images and extract key information
- Record punch-in data to daily note files
- Provide progress feedback and suggestions

This skill NEVER:
- Automatically perform any action based on recognition results
- Upload images to external services (except VLM API)
- Access image resources that the user has not authorized
- Modify the user's goal plan (only record progress)

---

## Security & Privacy

**Data that stays local:**
- Structured results after recognition
- Record to daily notes or long-term memory and the contents of USER.md
- Check-in historical data

**This skill does NOT:**
- Share goal progress or check-in data to third parties
- Automatically publish check-in information to social platforms
- Access the user's other image resources

---

## Notes

1. **Privacy Protection**: Images and recognition results are only stored locally and will not be uploaded to the cloud (except for calling the VLM API for recognition)
2. **Accuracy**: The content recognized by VLM is for reference only and may be biased due to blurred writing, picture quality, etc.
3. **Confirm in time**: It is recommended that users confirm the recognition results in time after recording. If there are any deviations, they can be corrected manually.
4. **Target type recognition**: The system will automatically determine the target type based on the image content. If there is an error, you can manually adjust it.
5. **Progress estimate**: The progress percentage is estimated based on the picture content and may not be accurate. Users are recommended to update manually regularly.

---

## Integration suggestions

### Works with SOUL.md

Integrate automated trackers into your daily goal management workflow:

```markdown
### 2. Intelligent Logging & Estimation

- When the user sends any image related to the target:
1. Automatically call auto-target-tracker to identify content
2. Extract key information and estimate progress
3. Immediately record it in your daily notes
4. Synchronously update the target progress of USER.md
```

### Works with HEARTBEAT.md

Included in the heartbeat check:

```markdown
## Daily summary
- 22:00 Automatically read all punch-in records today
- Generate goal progress reports
- Send to user
```