Skillget

plydb

mcp
Guvenlik Denetimi
Uyari
Health Uyari
  • License — License: Apache-2.0
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 9 GitHub stars
Code Gecti
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Gecti
  • Permissions — No dangerous permissions requested
Purpose
This tool acts as a universal database gateway for AI agents. It allows AI models to securely query live data across multiple databases (Postgres, MySQL) and flat files (CSV, Parquet) using standard SQL without requiring data movement.

Security Assessment
The tool inherently accesses sensitive data by design, acting as a bridge between an AI agent and your local or cloud databases. However, it defaults to read-only access, which mitigates the risk of accidental data modification or deletion by the AI. The automated code scan found no dangerous patterns, no hardcoded secrets, and no dangerous permission requests. Because it acts as a controlled access point, it limits the blast radius of what an agent can do compared to giving it direct database access. Overall risk is rated as Medium, primarily because connecting any tool to live production databases always requires careful configuration and oversight.

Quality Assessment
The project is very new and currently has low community visibility with only 9 GitHub stars, meaning it has not been broadly battle-tested. Despite its recent creation, it is actively maintained, with the last code push occurring today. The repository is properly licensed under Apache-2.0, making it safe for commercial and open-source use.

Verdict
Use with caution: the code is clean, licensed, and defaults to read-only access, but its low community adoption means you should monitor its queries closely when connecting it to sensitive production environments.
SUMMARY

PlyDB securely bridges the gap between your AI agents and your fragmented data sources. It provides a single controlled access point for AI agents to query databases and flat files, such as Postgres, MySQL, SQLite, DuckDB, CSV, Excel, Google Sheets, and Parquet, with SQL, wherever the data lives. Integrate via CLI or MCP.

README.md

PlyDB: The Universal Database Gateway for AI Agents

Real-time conversational analytics with zero data movement. Bridge your AI to
live sources without the ETL tax. Immediate insights, where your data lives.

PlyDB is a secure, unified access point for AI agents to query data in-place.
From SQL databases like Postgres and MySQL, flat files like CSV and Excel, or
cloud sources like Google Sheets and S3, PlyDB can query across them all with
standard SQL - no data warehouse necessary.

PlyDB is:

  • Simple: Deploy in minutes on a local machine with no additional
    infrastructure. Connect your agents to live data without the friction of
    building ETL pipelines.
  • Secure: Control which data sources your agents can access. Read-only by
    default.
  • Versatile: Query across different data sources
    through a single interface. Integrate your agents via
    CLI or
    Model Context Protocol (MCP).

Demo

PlyDB can connect with a wide variety of data sources
and integrate with any AI agent that supports
CLI tools or
MCP.

Here is a demo with Claude Desktop + PlyDB + Revenue Data:

Where does PlyDB fit in?

 ┌─────────────────┐ SQL ┌──────────────────────────────────────┐
 │     AI Agent    │────▶│               PlyDB                  │
 │(Claude, ChatGPT,│◀────│                                      │
 │ etc.)           │     │  ┌─────────────────────────────────┐ │
 └─────────────────┘     │  │          Query Engine           │ │
                         │  │   (Access control, Planning,    │ │
                         │  │    Optimization & Execution)    │ │
                         │  └───────────────┬─────────────────┘ │
                         │                  │                   │
                         │     ┌────────────┼────────────┐      │
                         └─────┼────────────┼────────────┼──────┘
                               ▼            ▼            ▼
                          ┌──────────┐ ┌──────────┐ ┌──────────┐
                          │PostgreSQL│ │  MySQL   │ │ S3/Local │
                          │ Database │ │ Database │ │  Files   │
                          └──────────┘ └──────────┘ └──────────┘

Why PlyDB?

  • Agentic Data Analysis: Unleash the full potential of AI agents by allowing
    them to write sophisticated SQL and perform complex data analysis
    autonomously. Agents can use either the PlyDB CLI or MCP server to discover
    tables, inspect schemas, and understand your data's semantics.
  • Zero ETL (Query In-Place): Eliminate the need for expensive and brittle
    ETL pipelines. PlyDB lets your agents query your data exactly where it lives -
    whether it's a production database, a cloud-hosted spreadsheet, or a data
    lake.
  • Read-Only Guardrails: PlyDB is a read-only "look, don't touch" system by
    default. Your AI can analyze information and find patterns, but it cannot
    delete, edit, or alter your original records unless you explicitly allow it
    to.
  • Cross-Source Queries: Join tables across MySQL, PostgreSQL, CSV, and more
    in a single query.
  • Operational Simplicity: Designed to be up and running in minutes without
    additional infrastructure dependencies.
  • Open Source & Extensible: Built on an open-source foundation, PlyDB
    ensures transparency, security, and no vendor lock-in.

Example Use Cases

When your AI agent has a secure, real-time view of your data, it evolves from a
chatbot into a Strategic Partner — one that can answer complex questions
about your business the moment you ask them.

DevOps

Prompt: "Review our app logs in S3 for errors from the past week.
Cross-reference with our codebase and database replica to identify affected
customers and diagnose root causes. Open PRs with fixes for the most severe
issues, and draft a summary our PM and CSM teams can use."

Strategic Sales & Retention

Prompt: "Analyze our top 20 accounts by revenue. Cross-reference their
support tickets with their recent product usage. Generate a churn-risk
dashboard and draft personalized 'Value Review' emails for the three accounts
with the lowest activity."

Marketing Performance & Optimization

Prompt: "Look at our ad spend across Google and Facebook, then compare it
with our actual transaction data. Create a chart showing the ROAS trend over
the last 90 days and identify which specific campaign we should move budget into
to maximize next month's yield."

Revenue Operations (RevOps)

Prompt: "Audit our active seat counts against our signed contracts in the
Google Sheet. Identify all overages, calculate the total unbilled revenue,
and build a summary table that the billing team can use to issue invoices."

Executive Insights

Prompt: "I need a high-level summary of our business health. Pull the MRR
from the CRM, the infrastructure costs from our logs, and the headcount from the
HR spreadsheet. Build a financial health dashboard and suggest three areas
where we can improve our operating margin."

Genomics & Bioinformatics

Prompt: "Query the variant_calls table to find the top 10 most frequent SNPs
found in samples labeled as 'resistant' to Penicillin, excluding variants found
in the 'control' group."

Public Health & Epidemiology

Prompt: "Calculate the rolling 12-month average of ER admissions for asthma
by zip code".

Supported Data Sources

PlyDB abstracts the complexity of different storage formats into a single
relational view:

Category Supported Sources
SQL Databases PostgreSQL, MySQL, SQLite, DuckDB
File Formats CSV, JSON, Parquet, Excel (.xlsx)
Object Storage S3
SaaS Google Sheets
Data Lake (planned) Apache Iceberg, Delta Lake

Installation

Quick install (macOS / Linux)

curl -fsSL https://raw.githubusercontent.com/kineticloom/plydb/main/install.sh | sh

Quick install (Windows — PowerShell)

irm https://raw.githubusercontent.com/kineticloom/plydb/main/install.ps1 | iex

Options

Variable Description Default
PLYDB_INSTALL_DIR Where to place the binary ~/.local/bin
PLYDB_VERSION Version tag to install (e.g. v0.1.0) latest

Manual download

Pre-built binaries for all platforms are available on the
Releases page.

Build from source

make build

AI agents + PlyDB via MCP

Deciding between MCP vs CLI? See the FAQ
for recommendations.

AI agents can connect to PlyDB via MCP.

Install PlyDB (we recommend using the quick install process).

Then follow your specific agent's instructions for configuring MCP:

AI agents + PlyDB via CLI Agent Skill

Deciding between MCP vs CLI? See the FAQ
for recommendations.

AI agents can also use PlyDB directly via the plydb CLI when provided context
on how to do so via an Agent Skill.

Install PlyDB (we recommend using the quick install process).

Then download the Agent Skill bundle
(plydb_skill.zip) and follow
your specific agent's instructions for installing skills:

Configuring data sources

Pro tip: If you have the
PlyDB Agent Skill installed, you can
ask your agent to work with you on data source configuration instead of
writing a config file manually.

Data sources are configured via the
PlyDB config file.

You can configure more than one type of data source in a config file, depending
on your needs, and query across all of them.

To guide your AI agent's understanding of the semantics of your data, PlyDB can
automatically scan your data sources and provide your AI agent with
semantic context - schema,
tables, columns, and comment metadata. You can further enrich this context by
overlaying
your own descriptions or AI-generated annotations.

Examples of configuring data sources:

FAQ

Contributing

We love contributions! However, before getting too deep into implementation,
please first check our Roadmap and start a discussion so we can
align on a direction.

Features to improve PlyDB for individuals can and should be a part of this open
source version, while features aimed to improve usage at scale, as in an
enterprise, should be reserved for the proprietary version. Doing so helps keep
the project sustainable. If you aren't sure where a feature fits, feel free to
open a discussion first!

See CONTRIBUTING.md for details.

License

This project is licensed under the Apache License Version 2.0. See the
LICENSE file for details.

All code contributed prior to 02/23/2026 is also licensed under Apache License
Version 2.0.

Yorumlar (0)

Sonuc bulunamadi