YashCh05

Dash Sales Agent — AI skill for Claude Code

AI community

Agentic AI sales dashboard for General Trade.

How to install Dash Sales Agent

This entry records only its repository, not the path inside it, so there is no exact command to give. Open YashCh05/dash-sales-agent and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Dash Sales Agent does

Agentic AI sales dashboard for General Trade. Cleans messy CSV exports into a DuckDB warehouse, then serves live KPIs, custom charts, natural-language querying, and anomaly detection. FastAPI + vanilla JS with a pluggable Claude/Gemini/HuggingFace backend.

Alternatives in AI

Engineering Notebook — A Bun CLI that ingests Claude Code and Codex session transcripts, generates LLM-powered daily summaries, and s 320 ★
Dataverse Skills — Microsoft Dataverse skills for AI coding agents 209 ★
Ppt Agent Skill — 🎨 World-class AI presentation generator · 26 styles · 18 charts · benchmarked against Linear/Anthropic/Stripe 139 ★

README

**DASH** turns raw, inconsistent General Trade sales exports into a clean queryable warehouse, a live dashboard, and a natural-language analyst — without ever letting an LLM write or execute code against your data.

Drop in a messy CSV. DASH works out what every column means, cleans it, files it away, and then sits on the dashboard writing chart captions, answering questions in plain English, and flagging numbers that look wrong.

DASH dashboard

Why this exists

General Trade distribution data is painful in three specific ways, and DASH is built around each one:

Problem	What DASH does about it
Every export is different. Column names shift between ERP dumps. Three date formats turn up in one column. Names arrive as `KRISHNA_TRADING_CO`.	One LLM call classifies every column at once against a canonical schema, then deterministic Python does the actual cleaning.
Primary ≠ Secondary. What the company invoices to distributors and what distributors actually sell to retailers are different numbers, and nobody reconciles them.	Both are modelled as first-class tables with a built-in variance query that surfaces distributor inventory pile-up.
Every question needs an analyst. "Which produc