# Dtcat

> View and analyze tabular data files (CSV, TSV, Parquet, Arrow/Feather, JSON, NDJSON, Excel) using dtcat. Use when the user asks to open, view, inspect, read, or analyze a data file in any of these formats, or when you encounter such a file that needs to be examined. Also use when converting between formats.

- **Type:** Skill
- **Install:** `agentstack add skill-louloulibs-claude-skills-dtcat`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [LouLouLibs](https://agentstack.voostack.com/s/louloulibs)
- **Installs:** 0
- **Category:** [Data & Analytics](https://agentstack.voostack.com/c/data-and-analytics)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [LouLouLibs](https://github.com/LouLouLibs)
- **Source:** https://github.com/LouLouLibs/claude-skills/tree/main/skills/dtcat

## Install

```sh
agentstack add skill-louloulibs-claude-skills-dtcat
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# dtcat — Tabular Data File Viewer

View and analyze tabular data files at the command line. Supports CSV, TSV, Parquet, Arrow/Feather, JSON, NDJSON, and Excel. Outputs structured, LLM-friendly markdown.

## Quick Reference

```bash
# Auto-detect format and show overview
dtcat data.csv
dtcat data.parquet
dtcat data.arrow
dtcat data.json
dtcat report.xlsx

# Column names and types only
dtcat data.parquet --schema

# Summary statistics per column
dtcat data.csv --describe

# File metadata (size, format, shape, sheets)
dtcat report.xlsx --info

# View a specific sheet (Excel only)
dtcat report.xlsx --sheet Revenue

# First/last N rows
dtcat data.csv --head 10
dtcat data.csv --tail 10

# Show all rows (override adaptive limit)
dtcat data.csv --all

# Random sample of rows
dtcat huge.parquet --sample 20
dtcat huge.parquet --sample 50 --csv

# Raw CSV output for piping
dtcat data.parquet --csv

# Convert between formats
dtcat data.csv --convert parquet -o data.parquet
dtcat report.xlsx --sheet Revenue --convert csv -o revenue.csv
dtcat data.parquet --convert ndjson              # text formats go to stdout

# Override format detection
dtcat data.txt --format csv

# Skip metadata rows above header
dtcat data.csv --skip 2
```

## Supported Formats

| Format | Extensions | Auto-detected |
|--------|-----------|---------------|
| CSV | .csv | Yes (magic + extension) |
| TSV | .tsv, .tab | Yes |
| Parquet | .parquet, .pq | Yes (PAR1 magic) |
| Arrow/Feather | .arrow, .feather, .ipc | Yes (ARROW1 magic) |
| JSON | .json | Yes ([ prefix) |
| NDJSON | .ndjson, .jsonl | Yes ({ prefix) |
| Excel | .xlsx, .xls, .xlsb, .ods | Yes (PK/OLE magic) |

Detection priority: `--format` flag > magic bytes > file extension.

## Default Behavior

- **50 rows:** shows first 25 + last 25 rows
- **Excel, multiple sheets:** lists all sheets with schemas (use `--sheet` to pick one)

## Flags

| Flag | Purpose |
|------|---------|
| `--format ` | Override format detection (csv, tsv, parquet, arrow, json, ndjson, excel) |
| `--sheet ` | Select sheet by name or 0-based index (Excel only) |
| `--skip ` | Skip first N rows before header |
| `--schema` | Column names and types only |
| `--describe` | Summary statistics (count, mean, std, min, max, median, unique) |
| `--head ` | First N rows |
| `--tail ` | Last N rows |
| `--all` | Show all rows (override adaptive limit) |
| `--sample ` | Randomly sample N rows |
| `--csv` | Output as CSV instead of markdown |
| `--info` | File metadata (size, format, shape, sheets) |
| `--convert ` | Convert to format (csv, tsv, parquet, arrow, json, ndjson) |
| `-o ` | Output file path (required for binary formats with --convert) |

## Exit Codes

| Code | Meaning |
|------|---------|
| 0 | Success |
| 1 | Runtime error (file not found, corrupt) |
| 2 | Invalid arguments |

## Workflow

1. Run `dtcat ` to see the overview (schema + sample data)
2. For Excel multi-sheet files, pick a sheet with `--sheet`
3. Use `--describe` for statistical analysis
4. Use `--head`/`--tail` to zoom into specific regions, `--sample` for random rows
5. Use `--csv` when you need to pipe data to other tools
6. Use `--convert` to transform between formats (e.g., CSV to Parquet)

## Mutual Exclusivity

- `--schema`, `--describe`, `--info`, `--csv`, `--convert` are mutually exclusive with each other
- `--sample` is mutually exclusive with `--head`, `--tail`, `--all`
- `--convert` is mutually exclusive with all display flags

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [LouLouLibs](https://github.com/LouLouLibs)
- **Source:** [LouLouLibs/claude-skills](https://github.com/LouLouLibs/claude-skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-louloulibs-claude-skills-dtcat
- Seller: https://agentstack.voostack.com/s/louloulibs
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.