# pipeverify

> pipeverify is an open-source MCP server for analytics and data. Compare data tables before and after a pipeline change and get PASS, REVIEW or BLOCK. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, self-hosted, macOS, Linux, Windows
- **Repository:** https://github.com/akshaysharma1088/pipeverify
- **MCP Registry name:** io.github.akshaysharma1088/pipeverify
- **Categories:** [analytics and data](https://ai-powered.com/categories/analytics-and-data), [developer tools](https://ai-powered.com/categories/developer-tools)
- **Listing updated:** 2026-11-18

## About

A pipeline change can finish on time, pass its tests and keep the schema intact while quietly dropping rows, duplicating keys or changing what a column means. pipeverify compares two snapshots, before and after, and optionally against the raw source, then returns a verdict: PASS (merge), REVIEW (something changed that tests would not catch) or BLOCK (the change damages the data; CI fails by default).

Checks include schema changes, row counts, missing and new keys, duplicate keys from join fan-out, null rates, changed values per column and column totals. Defaults differ by severity, for example BLOCK at a 10% row change. Sources can be CSV, Parquet, JSON or Excel files, DuckDB tables or queries, dbt models, Snowflake tables or any SQLAlchemy database. It never runs dbt and writes nothing, and a Snowflake run issues one SELECT per snapshot, so a read-only role is enough.

A YAML config can mark columns expected to change, ignore noisy ones and set thresholds. It also ships as a GitHub Action and an MCP server for agents. The README points to a benchmark where production-style monitors approved every harmful fix made by LLM agents.

## Features

- **Three-level verdict:** PASS, REVIEW or BLOCK with CI exit codes.
- **Key and row checks:** Detects lost rows, missing keys and duplicate-key fan-out.
- **Value and total checks:** Flags changed values, null creep and shifted column totals.
- **Many sources:** Files, DuckDB, dbt models, Snowflake and SQLAlchemy databases.
- **Configurable expectations:** Mark expected column changes and set thresholds in YAML.
- **Read-only operation:** Reads snapshots and writes nothing to your warehouse.

## Good for

- Gate a dbt pull request on data equality
- Verify an AI-written pipeline fix
- Catch a join fan-out that tests miss
- Detect a unit change in an upstream feed
- Run before-and-after checks in CI

## Install

### Claude Code (Terminal)

```sh
claude mcp add pipeverify -- uvx pipeverify
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "pipeverify": {
      "command": "uvx",
      "args": [
        "pipeverify"
      ]
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "pipeverify": {
      "command": "uvx",
      "args": [
        "pipeverify"
      ]
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "pipeverify": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "pipeverify"
      ]
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "pipeverify": {
      "command": "uvx",
      "args": [
        "pipeverify"
      ]
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "pipeverify": {
      "command": "uvx",
      "args": [
        "pipeverify"
      ]
    }
  }
}
```

## FAQ

### What verdicts does it return?

PASS, REVIEW or BLOCK; by default only BLOCK fails the CI check.

### Does it modify my warehouse?

No. It is read-only and Snowflake runs use one SELECT per snapshot.

### Which tools does it work with?

SQL, Spark, dbt, Airflow or an AI agent's fix, as long as it produces tables.

### How do I install it?

pip install pipeverify, then run pipeverify compare on two snapshots with a key.

### Is pipeverify free?

Yes. pipeverify is open source, so you can use and self-host it for free.

### How do I add pipeverify to Claude Code?

Run this command in your terminal: claude mcp add pipeverify -- uvx pipeverify

### What are the best alternatives to pipeverify?

The closest alternatives to pipeverify are REA, Context7, Chrome DevTools MCP, Codebase Memory and GitHub.

## Where this comes from

- Description: written from the project's README.
- Install steps: from the official MCP Registry entry.

## Alternatives

- [REA](https://ai-powered.com/mcp/rea): REA is an open-source MCP server for developer tools.
- [Context7](https://ai-powered.com/mcp/context7): Context7 is a freemium MCP server for developer tools.
- [Chrome DevTools MCP](https://ai-powered.com/mcp/chrome-devtools-mcp-2): Chrome DevTools MCP is an open-source MCP server for browser automation.
- [Codebase Memory](https://ai-powered.com/mcp/codebase-memory): Codebase Memory is an open-source MCP server for knowledge and memory.
- [GitHub](https://ai-powered.com/mcp/github-2): GitHub is an open-source MCP server for developer tools.
- [Serena MCP: the IDE for your agent](https://ai-powered.com/mcp/serena-mcp-the-ide-for-your-agent): Serena MCP: the IDE for your agent is an open-source MCP server for developer tools.
- [awesome-mcp-servers](https://ai-powered.com/mcp/awesome-mcp-servers-2): awesome-mcp-servers is an open-source MCP server for developer tools.
- [XERJ](https://ai-powered.com/mcp/xerj): XERJ is an open-source MCP server for knowledge and memory.

---
Source: https://ai-powered.com/mcp/pipeverify
