AI Agent Hub
Back to plugins
⚙️

dsh-data-cleaning-agent

Workflow Updated 2026.09.05

Run the following command in DeepSeek Harness:

dsh plugin install duhu2000/dsh-data-cleaning-agent

Paste the following prompt into your AI chat to install this plugin:

Install this plugin in DeepSeek Harness. Source: https://github.com/duhu2000/dsh-data-cleaning-agent

About this plugin

Enterprise name lists rarely arrive clean: phone numbers in inconsistent formats, blank or negative amount columns, missing required fields, and duplicated rows scattered throughout. dsh-data-cleaning-agent tackles this recurring pain point out of the box. Upload a CSV, XLSX, or JSON file and the local engine trims whitespace, normalizes phone numbers, drops rows with missing required values or negative amounts, removes duplicates, fills safe defaults, and produces a column overview with amount distribution. The result is a clean spreadsheet export, with no scripting required.

When deeper company context is needed, the plugin connects to the Qichacha (QCC) MCP to enrich records across business registration, ownership structure, historical filings, risk flags, intellectual property, and operational metrics. QCC calls follow a bring-your-own-account model: each customer consumes their own subscription quota. The plugin ships no developer keys, proxies no payments, and never auto-selects among multiple candidate companies. Every paid call requires explicit user confirmation, keeping billing and data sovereignty firmly in the customer's hands.

Inside DeepSeek Harness, the plugin opens a dedicated data-cleaning session with five capability buttons below the input bar (upload, clean and enrich, match and verify, profile, task history) and a right-hand workspace that walks through the full five-step loop from data upload to clean export. It supports pasting screenshots to extract company names, renders live tool-status cards in the conversation, and persists result artifacts across host restarts. Ideal for data analysts, sales and BD teams, risk and compliance staff, or anyone who wants natural-language-driven spreadsheet hygiene instead of manual scripting.

Use Cases

  • Upload a client-provided CSV of company names, clean phone and amount columns, and export a tidy spreadsheet
  • Batch-enrich company records with business registration, ownership, and risk data via Qichacha MCP
  • Paste a screenshot of company names, let OCR extract them, and feed them into the cleaning workflow

Best For

  • Data analysts and ETL engineers who want to cut manual spreadsheet-cleaning time
  • Sales, BD, and compliance teams needing business and risk profiles of company lists
  • DeepSeek Harness developers who want natural-language-driven table governance