A local, zero-dependency CSV, TSV, and JSONL data cleanup and inspection toolkit built entirely on the Python 3 standard library.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install clean-csv-toolkit
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install clean-csv-toolkit using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is clean-csv-toolkit?
The clean-csv-toolkit is a lightweight, secure utility collection engineered for AI agents to inspect, validate, and manipulate tabular data files without loading entire datasets into memory. It operates completely offline with no remote API or LLM calls, making it an excellent choice for privacy-centric environments utilizing Openclaw Skills.
By leveraging the Python 3 standard library rather than heavy frameworks like pandas, this tool processes files exceeding 100,000 rows in less than a second. It handles parsing, schema compliance, deduplication, merging, and transformations cleanly and safely, protecting workflows from shell injection vulnerabilities and memory exhaustion.
clean-csv-toolkit Use Cases
- Data Profiling: Quickly inspect unknown data drops to detect file encoding, column types, missing values, and row counts.
- Schema Validation: Implement automated data sanity checks within automated CI pipelines by verifying files against strict JSON schemas.
- Data Cleaning & Enrichment: Deduplicate entries, filter records via a safe expression language, and compute derived columns without risk of code execution.
- Data Translation: Seamlessly convert datasets among CSV, TSV, JSON Lines, JSON arrays, and GitHub-Flavored Markdown tables.
How clean-csv-toolkit Works
- Inspection and Dialect Auto-Detection: The toolkit checks the target file's structural metadata, establishing its dialect (CSV, TSV, or JSONL) and character encoding.
- Streaming and Processing: Actions like filtering, heading, and trailing stream data row-by-row using memory-efficient structures such as reservoir sampling and ring buffers.
- Safe Parsing and Calculation: Transformations and row selections pass through a hand-rolled tokenizer and recursive-descent parser, eliminating any reliance on dangerous Python
evalroutines. - Structured Format Output: Processed data is exported cleanly into the specified format, outputting exact exit codes (0 for success, 1 for failures/differences, 2 for bad arguments) to ensure continuous shell pipeline usability.
clean-csv-toolkit Setup
Verify that Python 3 is installed and ready within your environment:
bash scripts/check_deps.sh
No additional pip configuration or third-party packages are required. The ecosystem relies entirely on standard libraries.
clean-csv-toolkit Data Schema & Taxonomy
The toolkit interacts dynamically with a configurable schema file for validation tasks. It parses JSON rules structured as follows:
| Property | Type | Description |
|---|---|---|
required_columns |
Array | A list of headers that must be present in the document. |
columns |
Object | Map of column names containing rules such as type, min, max, regex, and unique. |
Validation Error Architecture
When validation fails, structured reports or JSON outputs outline rows with issues using the following taxonomy:
bad_type: Value does not match the inferred or declared datatype.regex_mismatch: Value fails to correspond with the expected regular expression pattern.below_min/above_max: Numeric indicators fall outside bounds.duplicate_unique: A constraint violation on a column requiring distinct rows.
clean-csv-toolkit Advanced Features
- Safe Transform Matrix Engine: Add calculated metrics via expressions like
profit = revenue - costusing built-in deterministic operations (coalesce,year,strip,replace). - Streamed Relational Joins: Merge separate files on single or multiple keys supporting
inner,left,right, andouterstrategies while controlling memory overhead. - Aggregation Cross-Tabs: Construct multi-variable group-by matrices and wide pivot reports with numeric-aware sorting directly inside terminal prompts.
- Reservoir Sampling: Pull perfectly random, reproducible sample subsets from massive data streams using arbitrary
--seedvalues to preserve analytical integrity.
SKILL.md
Loading
Related Openclaw Skills
WiseDiag Calories
An AI-powered food item identification and calorie estimation tool that processes local food photos or image URLs to return nutritional insights.
WiseDiag Medical Report Interpretation (wisediag-report)
An AI-powered medical report interpreter that recognizes text, highlights abnormal indicators, and provides health advice from local files or image URLs.
WiseDiag Skin
An AI-powered local skin disease analysis tool that securely processes local photos using WiseDiag AI and streams diagnostic references.
wisediag-checkup
Analyze and interpret medical checkup reports from PDFs or local files with WiseDiag's advanced AI health analysis.
Audio Transcribe ZH
An optimized offline speech-to-text script for rapid Chinese transcription powered by FunASR Paraformer.
thesis-review-pro
An expert AI coding agent skill designed to automate structured academic thesis evaluation and generate official Word-formatted review reports based on historical peer-review expertise.
Featured*
