Install Data Cleaner MCP

Clean, sanitize, and structure unstructured data using our hosted MCP server. No local setup or keys required for public endpoints.

Quick Setup

Paste this prompt into any AI agent to auto-install:

Install the Agentic Data Cleaner MCP server using URL: https://dc-mcp-26zy.onrender.com/mcp

01. CLIENT_CONFIG

v2.0.0
https://dc-mcp-26zy.onrender.com/mcp 
{

  "mcpServers": {

    "agentic-data-cleaner": {

      "serverUrl": "https://dc-mcp-26zy.onrender.com/mcp"

    }

  }

}

02. ANTIGRAVITY_SETUP

# Using antigravity to bridge to the cleaner

{

  "bridge": "antigravity",

  "target": "https://dc-mcp-26zy.onrender.com/mcp"

}

03. LARGE_FILES

Handling payloads over 25MB.

Open Temporary Pipe

25MB Cap / 1-Hour Purge Window

DEBUGGING

401_UNAUTHORIZED

Verify endpoint URL is exact. Hosted server needs no key. A 401 means you sent a wrong Authorization header — remove it.

CONNECTION_TIMEOUT

Server might be sleeping. Attempt a manual ping to dc-mcp-26zy.onrender.com to wake.

AI_AUTO_INSTALL

Paste this to any AI assistant — it connects to the hosted server and proves a real clean. No git, no install.

# Auto-connect prompt — paste this to any AI assistant

Copy everything below the line and paste it into Claude, ChatGPT, Cursor,
VS Code, Opencode, Antigravity or any other AI assistant. No downloads, no
install, no git: it connects the assistant to the hosted MCP server and
proves it works.

---

Connect me to the Agentic Data Cleaner MCP server. Do everything yourself and
stop only when a real file has been cleaned through the server:

1. **Configure my client:** add a remote MCP server entry pointing at
   `https://dc-mcp-26zy.onrender.com/mcp` (Streamable HTTP, no Authorization
   header needed — the server admits keyless visitors). Use the exact config
   shape my assistant expects (Antigravity: `serverUrl`; Claude/Cursor/opencode:
   `url`). Show me the snippet, tell me exactly which file to save it in for
   the assistant I am using right now, and remind me to restart/reload the
   assistant afterwards if that is required for MCP changes to take effect.

2. **Prove the handshake:** `POST https://dc-mcp-26zy.onrender.com/mcp` with
   header `Accept: application/json, text/event-stream` and body
   `{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}` must return
   all 10 tools: clean_dataset, scan_dataset, get_cleaning_plan,
   apply_cleaning_plan, apply_cleaning_action, preview_dataset, export_dataset,
   create_upload_url, delete_dataset, list_cleaning_actions.

3. **Prove a real clean:** call `clean_dataset` with
   `{"csv_text":"name,salary\n\"  Alice \",50000\nBob,N/A\n","filename":"t.csv"}`.
   The result must show safe fixes applied (whitespace trimmed, N/A nulled),
   risky items held under `needs_your_decision`, and a `cleaned_data` download
   link. Download it and show me the cleaned content: `Alice,50000` and `Bob,`
   with the salary empty (null preserved, nothing invented).

4. **Teach me the loop:** in one short paragraph explain that I paste or point
   at messy data, the assistant calls `clean_dataset`, safe fixes apply
   automatically, and anything destructive (deleting rows/columns, filling
   values, masking personal data) comes back for my explicit approval via
   `apply_cleaning_plan(approve_ids=[...])`. Originals are never modified.

Rules: every check must genuinely pass against the live server — show me the
evidence, not claims. If a call fails twice (note: the free Render host sleeps
when idle, so a first slow call is normal — retry), stop and show me the exact
error and what you tried. Docs: https://dc-mcp-26zy.onrender.com/install

---

*Agentic Data Cleaner by Ishan Makkar — https://github.com/iishanmakkar/dc-mcp*