# Report Format — CSV Data-Quality Profile

Render the profiler's JSON into exactly this Markdown structure. Do not add, remove, or reorder sections.

# Data Profile: <file name>

**Verdict:** <one of the three verdict lines below>

## Overview
| Metric | Value |
|---|---|
| Rows profiled | <rows> (+ " (truncated at 100,000)" if truncated) |
| Columns | <number of columns> |
| Duplicate rows | <duplicate_rows> |
| Ragged rows | <ragged_rows> |

## Columns
| Column | Type | Nulls | Unique | Stats |
|---|---|---|---|---|
| <name> | <type> | <nulls> | <unique> | <stats cell — see rule below> |

## Issues
- <one bullet per detected issue; write "None detected." if empty>

# Cell and verdict rules

Stats cell:

• numeric columns → min=<min>, max=<max>, mean=<mean>, median=<median> plus , outliers=<n> when the key is present and > 0
• string columns → top: <value> (<count>), <value> (<count>), ... from top_values
• boolean/date/empty columns → —

Issue bullets (include each only when its condition is true):

• nulls > 0 in a column → "<column> has missing values (%)"
• duplicate_rows > 0 → " duplicate rows"
• ragged_rows > 0 → " rows have an inconsistent number of fields"
• outliers present and > 0 → "<column> has outliers (>3σ from mean)"
• encoding_fallback is true → "file is not valid UTF-8; profiled using latin-1 fallback"
• truncated is true → "profile limited to the first 100,000 rows"
• rows == 0 → "file contains no data rows"

Verdict line (pick exactly one, in this priority order):

1. rows == 0 OR ragged_rows > 0 → ⚠️ Needs attention — <short reason>
2. any other issue bullet present → 🟡 Usable with caveats — <n> issue(s) found
3. no issue bullets → ✅ Clean — no data-quality issues detected