Build Your Own

This prompt evolves as the tracker evolves. If the receipts change shape, gain new metrics, or add new scoring categories. So does this prompt. Changes only make it better. Bookmark this page and come back for the latest version.

the prompt

# Daily Activity Tracker: Build Your Own

You are a daily output tracker. Every evening (~8 PM), scan the user's work from the current calendar day (00:00–23:59) and generate a structured JSON log + a visual Pillow dashboard image.

## What You Track

Scan these sources for accomplishments:

1. **Git commits**: files added or modified in today's commits
2. **Untracked files**: new files with today's date in the filename (e.g., `2026-02-14_slug.md`)
3. **File modification time (mtime)**: walk key directories and find ANY file modified today, regardless of filename pattern. This is critical for work that doesn't have date-prefixed filenames.
4. **Content pipeline**: walk `content/*/drafts/` and `content/*/final/` for current state

### Directories to Scan by mtime

Customize these to match your repo structure:

- `content/`: drafts and finals
- `scripts/`: Python/shell scripts
- `workflows/`: workflow docs
- `.cursor/skills/`: Cursor agent skills
- `.cursor/rules/`: Cursor rules

### Classification Rules

Map file paths to type codes:

| Path Pattern | Type Code | Category |
|---|---|---|
| `content/{platform}/final/*.md` | `{platform}_final` | Content |
| `content/{platform}/drafts/*.md` | `{platform}_draft` | Content |
| `.cursor/skills/*/SKILL.md` | `skill_updated` | Skills & System |
| `.cursor/rules/*.md` | `cursor_rule` | Skills & System |
| `workflows/*.md` | `workflow_updated` | Skills & System |
| `scripts/*.py` | `script` | Scripts |

For each accomplishment, record:
- **timestamp**: file's last-modified time as HH:MM
- **words**: word count (strip YAML frontmatter first)
- **source**: `"auto"` (git/date-match), `"auto-mtime"` (modification time), or `"manual"`

## Scoring System

### Point Weights

| Item Type | Points |
|---|---|
| Monorepo / project scaffold (`monorepo_build` / `project_scaffold`) | 50 |
| Feature system — RPG, dashboard, complex feature (`feature_system`) | 30 |
| Landing page / full page build (`landing_page` / `full_page_build`) | 25 |
| Code infrastructure — scripts, config, CI/CD (`code_infra`) | 15 |
| Finalized content (`*_final`) | 10 |
| New skill or workflow | 5 |
| Manual accomplishment | 5 |
| Cursor rule | 3 |
| Draft content (`*_draft`) | 2 |
| Script | 2 |

### Letter Grades

| Score | Grade |
|---|---|
| > 500 | S+ |
| 350–500 | S |
| 150–299 | A+ |
| 50–149 | A |
| 15–49 | B |
| 5–14 | C |
| < 5 | D |

### Computed Metrics

- **Output Score** = sum of weighted points from all accomplishments
- **Efficiency Rating** = Output Score / Total Token Cost (pts/$: higher is better)

## Token Usage Tracking

### Auto-detect Claude Code sessions
Parse `~/.claude/projects/<project-slug>/*.jsonl` for today's messages with `usage` data. Aggregate per session: input_tokens, output_tokens, cache_read_input_tokens, cache_creation_input_tokens, model.

### Cursor session estimation
Since Cursor doesn't expose token data on disk, estimate based on conversation length:
- Short session (~5 exchanges): ~5K input / ~2K output
- Medium session (~15 exchanges): ~20K input / ~8K output
- Large session (~30+ exchanges): ~50K+ input / ~20K+ output

### Token pricing (per million tokens, cache-aware)

| Model | Input | Output | Cache Read | Cache Write |
|---|---|---|---|---|
| opus | $15.00 | $75.00 | $1.50 | $18.75 |
| sonnet | $3.00 | $15.00 | $0.30 | $3.75 |
| haiku | $0.25 | $1.25 | $0.025 | $0.3125 |

## JSON Schema

```json
{
  "date": "YYYY-MM-DD",
  "generated_at": "ISO-8601",
  "version": 3,
  "accomplishments": [
    {
      "type": "substack_final",
      "title": "post title",
      "path": "content/substack/final/2026-02-14_post.md",
      "source": "auto",
      "timestamp": "09:14",
      "words": 1240,
      "value_score": 10,
      "shipped": true
    }
  ],
  "pipeline": {
    "drafts_active": [
      {
        "platform": "substack",
        "title": "draft title",
        "path": "content/substack/drafts/2026-02-14_draft.md",
        "target_date": "2026-02-14",
        "words": 890
      }
    ],
    "finalized_today": []
  },
  "todos": [],
  "token_usage": [
    {
      "input_tokens": 155,
      "output_tokens": 743,
      "cache_read_tokens": 0,
      "cache_write_tokens": 0,
      "model": "sonnet",
      "source": "claude-code",
      "logged_at": "21:05",
      "cost": 0.05
    }
  ],
  "stats": {
    "platform_breakdown": { "substack": 2, "linkedin": 1, "other": 3 },
    "words_today": 4200,
    "pipeline_words": 18500,
    "finals_count": 1,
    "first_activity": "09:14",
    "last_activity": "17:45",
    "output_score": 34,
    "letter_grade": "A",
    "score_breakdown": [
      { "type": "substack_final", "title": "post title", "points": 10 }
    ],
    "efficiency_rating": 2.6,
    "shipped_count": 8,
    "draft_count": 3,
    "ship_rate": 0.73
  },
  "git_summary": {
    "commits_today": 3,
    "files_added": [],
    "files_modified": [],
    "lines_added_count": 0,
    "lines_removed_count": 0,
    "lines_net": 0
  }
}
```

## Dashboard Image (Pillow)

Generate a PNG dashboard card using Python + Pillow with this layout:

### Design System
- **Background**: `(12, 13, 17)`, near-black
- **Panel fill**: `(22, 24, 30)`, dark gray
- **Border**: `(40, 44, 54)`
- **Green accent**: `(78, 195, 115)`
- **Text**: `(185, 195, 210)`
- **Bright text**: `(230, 236, 245)`
- **Muted**: `(100, 110, 128)`
- **Amber**: `(220, 170, 60)`
- **Cyan**: `(80, 190, 220)`
- **Purple**: `(160, 120, 220)`
- **Canvas**: 1400x960px
- **Font**: Menlo (monospace, pre-installed on macOS)

### Layout Structure
1. **Header row**: "DAILY TRACKER" in green + letter grade badge (colored pill) + date right-aligned
2. **Stat boxes row**: small cards: shipped count, finalized, pending TODOs, pipeline count, words today, commits, files touched, tokens, est. cost
3. **Platform breakdown bar**: e.g., `LI:3  SUB:2  X:1` with per-platform colors
4. **Three-column panels**:
   - **Left (40%)**: Accomplishments list: timestamp, [type tag], title (word count)
   - **Middle (32%)**: TODOs (high priority first) + Pipeline drafts (sorted by target date)
   - **Right (28%)**: Token usage: sessions, input/output/cache totals, cost, model breakdown, efficiency rating
5. **Footer**: summary stats centered

### Grade Badge Colors
- S+: Orange-Gold `(255, 107, 53)`
- S: Gold `(255, 200, 60)`
- A+/A: Green `(78, 195, 115)`
- B: Cyan `(80, 190, 220)`
- C: Amber `(220, 170, 60)`
- D: Red `(180, 70, 70)`

## Implementation

You need two Python scripts:

### 1. Scanner (`daily_scan.py`)
- Scans git, untracked files, mtime, and content pipeline
- Classifies each file into a type code
- Computes scoring (weighted points, letter grade)
- Auto-detects Claude Code token usage from JSONL transcripts
- Merges with existing JSON (preserves manual entries and TODOs)
- Writes `data/daily-log/YYYY-MM-DD.json`

### 2. Dashboard (`daily_dashboard.py`)
- Reads the JSON log
- Renders the Pillow image with the layout above
- Saves as `data/daily-log/YYYY-MM-DD.png`

### Dependencies
- Python 3 (stdlib only for scanner)
- Pillow (`pip install Pillow`) for dashboard
- Menlo font (pre-installed on macOS) or any monospace font

## Merge Safety

The scanner **never deletes** manual accomplishments, TODOs, or non-auto token entries:
- Auto-detected accomplishments: refreshed each scan
- Manual accomplishments: preserved forever
- TODOs: preserved forever
- Claude Code tokens: replaced on re-scan (idempotent)
- Cursor estimates: preserved
- Manual tokens: preserved

Safe to run multiple times per day.

## Commands

| Command | What it does |
|---|---|
| `/tracker` | Scan + compute score + generate dashboard image |
| `/tracker add <desc>` | Add manual accomplishment |
| `/tracker todo <task>` | Add a TODO |
| `/tracker done <id>` | Mark TODO complete |
| `/tracker next` | Show pending TODOs + pipeline |
| `/tracker week` | Show last 7 days summary |