Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/UnkleFunk/HouseMusicSwarm-/llms.txt

Use this file to discover all available pages before exploring further.

The Data Analyst runs all computation inside an isolated IPython kernel via the IPythonInterpreter tool. State persists across tool invocations within the same session, so the agent can load a dataset, clean it, run multiple analyses, generate charts, and save outputs — all in a continuous Python environment without re-importing libraries or reloading data between steps. This makes it practical for long analysis workflows that would otherwise require constant re-initialization.

​
Capabilities

Data analysis

Loads CSV and Excel files into pandas DataFrames, computes aggregations, identifies trends and anomalies, and surfaces KPIs relevant to the user’s business question.

Visualization

Generates charts with matplotlib, seaborn, and plotly. Saves outputs as PNG or SVG to ./mnt/outputs/. Loads generated images back with LoadFileAttachment to visually verify correctness before delivering.

Statistical modeling

Runs regression, clustering, time-series decomposition, and hypothesis tests using scipy, statsmodels, and scikit-learn. Results are always tied back to a concrete business question.

External platform data

Connects to Google Analytics, Stripe, Shopify, HubSpot, Salesforce, Google Sheets, BigQuery, Mixpanel, Amplitude, QuickBooks, and more via Composio. Uses ManageConnections + FindTools before executing any external call.

Shell operations

Uses PersistentShellTool for local filesystem operations — moving generated charts, reading environment variables, checking credentials, and renaming output files.

Web research

Uses WebSearchTool to look up API documentation, library references, or external benchmarks when needed during analysis.

​
Supported file formats

FormatNotes
CSV (.csv)Loaded directly with pandas.read_csv()
Excel (.xlsx, .xls)Loaded with pandas.read_excel() via openpyxl (xlsx) and xlrd (xls)
Structured data from external platformsFetched via Composio and converted to DataFrames in-kernel

​
Python environment

The IPythonInterpreter environment has all of the following libraries installed and available: 
# Data analysis
import pandas as pd
import numpy as np
import scipy
import statsmodels.api as sm
from sklearn.linear_model import LinearRegression   # scikit-learn

# Visualization
import matplotlib.pyplot as plt
import seaborn as sns
import plotly.express as px

# File handling
import openpyxl          # xlsx read/write
import xlrd              # xls read
import requests
from dotenv import load_dotenv
Charts and analysis outputs are always saved to ./mnt/outputs/ unless the user provides a different output path.
Charts and other output files are saved locally and the file path is included in the agent’s response. Use the path to open, share, or hand the file off to another agent (such as the Slides Agent or Docs Agent).

​
Workflow

1

Clarify the request

Identify the business question, required metrics, data source (file upload or external platform), and time period. Confirm any filters or segments needed.
2

Connect to data sources

For external platforms: run ManageConnections to check authentication, then FindTools to discover the right Composio tool, then fetch data inside IPythonInterpreter.For local files: load directly with pandas — no authentication required.
3

Analyze and visualize

Clean and transform data, calculate metrics, identify trends and anomalies. Generate charts and save them to ./mnt/outputs/. Load images back with LoadFileAttachment to verify formatting.
4

Deliver insights

Present concise findings tied to the original business question. Include file paths for every generated chart or output file. Call out data quality assumptions and limitations.

​
Example prompts

Analyze this CSV file and show me the key trends. Highlight any anomalies.

Create a bar chart of monthly revenue from this Excel file and save it as a PNG.

Run a linear regression on this sales dataset and tell me which features predict revenue best.

Connect to my Google Analytics and summarize last month's top traffic sources and conversion rates.

Build a dashboard with charts from my Stripe data — monthly MRR, churn rate, and new subscriber count.

​
Output format

A completed analysis always includes:
  • Scope and Sources — data sources, time period, metrics examined
  • Key Findings — 3–5 most important insights in plain language
  • What to Do Next — immediate actionable recommendations, prioritized by impact
  • Assumptions and Limits — data quality notes, missing information, confidence level
  • Follow-Up Actions — additional analysis worth running, data to track going forward

Build docs developers (and LLMs) love

Get started for freeTalk to us