About this skill

Problem It Solves

When you have local JSON, CSV, or JSONL records, such as task lists, personal datasets, or business exports, and want to see which themes recur, which values co-occur, and which entries deviate from the norm, pattern-miner gives you a local statistical analysis entry point. It does not call external APIs or upload files to the cloud, making it suitable for engineers who care about data visibility and privacy.

How It Works

The skill analyzes structured files you explicitly provide. Core capabilities include:
- Pattern recognition: extract recurring patterns from object fields, text themes, or value distributions.
- Association discovery: identify fields, tags, or values that frequently appear together, with support-style metrics.
- Anomaly detection: flag unusual items that deviate from common distributions or rules.
- Local reporting: results include confidence scores, support metrics, anomaly flags, and can be exported as JSON or CSV.

Supported input formats are JSON arrays of objects, header-based CSV, and JSONL with one JSON object per line. Configuration is typically stored in ~/.pattern-miner/config.json, and the analysis script is scripts/analyze.py. Runtime requires Python 3.8+ plus numpy, scikit-learn, and pandas for pattern statistics and anomaly checks.

Boundaries

It only reads files you explicitly specify, and does not access system logs, shell history, or OpenClaw session data. The data should be sufficiently structured; free-form text, binary files, or schemaless tables are not ideal direct inputs. It fits exploratory analysis of personal records, task lists, and local business exports, not production data warehousing, real-time stream processing, or complex modeling platforms.

Screenshots

Use Cases

Analyze CSV task exports to find recurring delay reasons and related fields.
Cluster JSONL behavior logs and flag entries that deviate from common paths.
Inspect a personal JSON note set to identify items and topics that co-occur.
Export a local JSON report for reviewing config value distributions and anomaly flags.

Best For

Data analysts processing local CSV exports who need offline association and anomaly checks.
Task managers organizing personal JSON notes who want recurring themes.
System operators auditing local JSONL logs who need anomaly flags.
Product analysts producing offline review reports with confidence scores.