Deploy and orchestrate hosted AI agents with Ash. Covers: creating an Ash client, deploying agents, managing sessions (create, pause, resume, end), sending messages, streaming SSE responses, sandbox isolation, multi-turn conversations, real-time text deltas, error handling, and working with files in agent workspaces. Works with TypeScript (@ash-ai/sdk) and Python (ash-ai-sdk) SDKs.

2026-03-0734

name	analyze
description	Analyze data by writing and running Python scripts. Can generate sample data, process CSVs, compute statistics, and create charts.
use_when	User asks to analyze data, compute statistics, process a dataset, create charts, or do quantitative analysis
allowed-tools	Bash(./scripts/), Bash(python3:), Bash(pip:*), Read, Write

Analyze

Run data analysis using Python scripts and the helper tools in scripts/.

Process

Understand what data to analyze — use provided files or generate sample data
Use ./scripts/csv-stats.sh <file> for quick CSV overview if applicable
Write a Python script for deeper analysis
Run the script and capture results
If charts are needed, use ./scripts/chart.py or write matplotlib code
Save the full analysis to analysis.md
Print the key findings

Available Tools

./scripts/csv-stats.sh <file> — Quick row count, column names, basic stats
./scripts/chart.py <csv> <x-col> <y-col> <output.png> — Generate a chart
Python 3 with standard library (csv, json, math, statistics, etc.)
Install packages with pip install if needed (pandas, matplotlib, numpy)