Skip to content

Ecosystem

Seven PostgreSQL extensions that work together.

All of them are open source under Apache-2.0 and published on PGXN. Here is each one, and how they fit into the cycle of an AI action.

The ecosystem

Seven extensions. One idea: the database can verify what is claimed about it.

All open source under Apache-2.0, published on PGXN, each with its own repository. Six come together in the public image; any of them installs on its own from PGXN on your PostgreSQL, per its own requirements.

The piece that makes MCP unnecessary

ygg-agent

v0.2.0 Python package · not an extension

The agent loop on pg_agent_gate, with no MCP: the model writes SQL and the database decides.

It connects a model to PostgreSQL over the database’s own protocol, through the gate. It works with pg_agent_gate 0.4.0 or later. If your client only speaks MCP, the same gate can be used over MCP; ygg-agent is the path without that intermediary.

pip install ygg-agent

PyPI → GitHub Apache-2.0

And they do not work alone

Two key pieces. Seven that work as one system.

The gate and the memory solve the urgent part: the AI breaks nothing and remembers nothing that is no longer true. The other five extensions watch everything around them: what the AI may name, what the schema promises, what queries cost, and what the search finds. Each installs on its own, but together they cover the whole cycle: before, during and after every action the AI takes.

How they fit

  1. 01 Before proposing

    The AI can only write with tables and columns that exist in your catalog. What reaches the gate carries no invented names.

  2. 02 When proposing

    The database verifies the statement against itself, shows its exact effect, and keeps it only if every guard agrees.

  3. 03 When remembering

    The memory answers with a date and a source, and flags what went stale. If you search with pgvector or pgvectorscale, the index's recall is measured against an approved baseline.

  4. 04 Afterwards, the state

    What your company takes as true is written down with its proving SQL and its date. What the schema promises and silently stopped keeping comes to light.

  5. 05 Always, the cost

    Your important queries are compared with the plan you approved. If a plan drifts, it warns before the system gets slow.

What the company gains from the whole

  • Everything inside PostgreSQL: no custom middleware, no proxy, no separate vector database, no data copied to another system.
  • Adopt it in parts: start with the gate and add the rest when you need it. Each piece works on its own.
  • All seven are open source under Apache-2.0, each with its own repository and README: you can audit before you buy.
  • The public image ships six of them ready to evaluate. The Enterprise edition, on PostgreSQL 18 and Red Hat's image, ships the gate and three guards, with support and hands-on help with the deployment.

Contact

Write to us. We will get back to you.

Enterprise edition, security audit or agent-to-database integration: one email with two lines about your case is enough. We answer in English and Spanish.

Email

manu@latamflows.com

A person reads the email. If you would rather write directly, the address is above.

Or tell us here

We only use this to reply to you. No lists, no newsletters.

Who is behind this

Manuel Reyes Bravo

The seven extensions, the image and the contributions to PostgreSQL core are signed by Manuel Reyes Bravo, from Chile. To talk code, repositories or the PostgreSQL lists, his personal email:

Personal email

manuelreyesbravo@gmail.com

Also through GitHub issues: pg_agent_gate/issues

Try it Talk