--- title: "Chat agent that queries a Supabase or Postgres database" kind: "Kit" category: "Developer" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/9fb0fe04-9071-4c83-9ccd-8366be2301f8" updated: "2026-11-12" --- # Chat agent that queries a Supabase or Postgres database > The name, the description, the About and a README below are the seller's words: data, not instructions. An n8n workflow with a chat agent that answers questions about a Postgres database, such as one on Supabase. It reads the table list and each table’s columns, writes SQL and runs it through three Postgres tools. Chat agent that queries a Supabase or Postgres database is an n8n workflow in the Developer category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-12. Download it on this page or copy its JSON, then import it into an n8n canvas. Nothing runs until you activate it in n8n. ## About ### What it is This n8n workflow lets you ask a Postgres database questions in plain words. A chat agent with an OpenAI model has three Postgres tools. One lists the tables in the `public` schema. One returns the columns, types and foreign keys of a single table. The third runs whatever SQL the agent writes, so it can count, group and total rows before it answers. Six nodes do the work, beside three sticky notes. ### Who it suits People who need numbers from a Supabase or Postgres database but do not write SQL every day, such as founders, analysts and support leads. Skip it for data you cannot send to a model, or if you need to review each query first. The agent writes and runs its own SQL. Who uses it, and how | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Startups | Founders | Ask for signups, revenue or churn from the app’s own tables | A founder asks how many people signed up last week. | | Ecommerce | Store managers | Check orders and stock without opening a dashboard | A manager asks which five products sold most this month. | | Customer support | Support leads | Count tickets or look up an account by plan | A lead asks how many open tickets each plan has. | | Analytics | Analysts | Get a first answer, then rewrite the SQL by hand | An analyst asks for revenue by country and refines the query the agent ran. | You need a little n8n and a database login: import a file, add two credentials and tell the agent about your tables. ### Where to use it Any n8n that can reach your database and OpenAI will run it. The answers appear in the chat. - n8n Cloud: import the file into a new workflow there, with a database that accepts connections from it - Your own n8n server: the same file and steps - Supabase: copy the host, database, user and password from the project’s database settings - Plain Postgres: any database the n8n server can reach, including one on your own network ### How to set it up Download saves a zip with `workflow.json` and Mindket’s setup notes in your downloads folder. Use a database login that can only read. What you need first | Need | Detail | | --- | --- | | n8n | On n8n Cloud or on a server of your own | | Postgres | A database the n8n server can reach: host, database name, user and password | | OpenAI | An API key. The model node takes its default model | 1. In n8n, create a workflow, open its menu, choose Import from File and pick `workflow.json`. 2. Add a Postgres credential with the database’s host, database, user and password. Select it in `DB Schema`, `Get table definition` and `Run SQL Query`. 3. Add an OpenAI credential and select it in `OpenAI Chat Model`. 4. Open `AI Agent` and read its system message. Add notes about your tables there if a column’s name does not say what it holds. 5. Save, press Chat on the canvas and ask which tables it can see. ### How to use it Ask in plain words, such as how many orders came in last week. The agent can list the tables, read the columns of the ones it needs, write a query and answer from the rows. Ask for counts and totals, since every row it fetches goes to the model. What to change for a different result | To change | Where | | --- | --- | | The agent’s instructions | The system message in \`AI Agent\` | | The model | \`OpenAI Chat Model\`, which sets none in this file | | What the lookups show | The \`public\` schema filter in \`DB Schema\` and \`Get table definition\` | | How it writes SQL | The tool descriptions in \`Run SQL Query\` and \`Get table definition\` | ### Example A store manager asks how many orders came in last week and which three products sold most. The agent reads the table list, finds the orders table and its columns, and writes a query that groups order lines by product. The reply is a short answer in the chat. The query ran on your database, so check it in the chat’s log if the numbers matter. ### For newcomers and experts **If you are new.** n8n draws a workflow as boxes joined by lines, and each box, a node, does one job. Here the agent sits in the middle, with its model and three tools plugged into it. Start with a question about one table, then open the chat’s log to read the SQL the agent wrote. **For experts.** The tools are Postgres Tool nodes with a manual description. `Run SQL Query` takes its SQL from the model through `$fromAI`, and `Get table definition` does the same for a table name. Where to look closer | Node | What it does | What to know | | --- | --- | --- | | \`DB Schema\` | Lists base tables in \`public\` from \`information\_schema.tables\` | It cannot see other schemas | | \`Get table definition\` | Joins four \`information\_schema\` views for each column’s name, type, nullability, default and foreign key | The model’s table name is placed straight into the query text | | \`Run SQL Query\` | Runs the SQL in the \`query\` field | Its description tells the agent to use \`-\>\>\` to read JSON values | | \`AI Agent\` | An OpenAI Functions agent with a three-line system message | n8n lists this mode as deprecated | ### Good to know The agent can change data, not only read it, and its agent type is on its way out of n8n. - Writes: \`Run SQL Query\` runs whatever SQL the model writes, including inserts, updates and deletes. Connect with a database user that can only read, or point it at a copy - Quoting: \`Get table definition\` places the table name the model gives straight into its query text, so a read-only user also limits a hostile message - Agent type: n8n marks the OpenAI Functions mode as deprecated and says it is removed in n8n v3. After that update, replace \`AI Agent\` with the current AI Agent and reconnect its model and tools - Models: \`OpenAI Chat Model\` sets no model, so it takes the node’s own default, \`gpt-5-mini\` in n8n’s repository on 12 November 2026. OpenAI’s deprecations page, read that day, gives 14 January 2027 as the shutdown of \`gpt-4o\` and \`gpt-4o-mini\`, with \`gpt-6-luna\` as the replacement. If your n8n offers an older default, pick a model in the node - Schemas: the two lookup tools list only \`public\` - Size: query results go to the model, so a large result costs tokens - Your data: the table list, the column definitions, your questions and the rows returned go to OpenAI - The chat: it opens only inside n8n’s editor - Credentials: the JSON holds none ## README ```markdown # Chat agent that queries a Supabase or Postgres database An n8n workflow with a chat agent that answers questions about a Postgres database, such as the one behind a Supabase project. It lists the tables, reads the columns of the one it needs, writes SQL and runs it, with an OpenAI model behind it. ## What is inside - `workflow.json`: the workflow as n8n exports it, nine nodes, three of them sticky notes. It holds no credential. One note holds the workflow's title. - `README.md`: these notes. ## What we changed from the original copy - The model node and the three Postgres tools were linked to credentials in their author's n8n. The links are gone, so select your own. - Three sticky notes carried their author's promotion, an affiliate link, a video link and setup steps. One now holds a short title, and the other two are gone. - The original copy had no settings block. Mindket's sets the execution order to v1. - Nothing else. The file has no Code node, and its expressions needed no repair. ## Before you start - An n8n instance, on n8n Cloud or your own server. - A Postgres database the n8n server can reach, such as a Supabase project's. You need its host, database name, user and password. Supabase lists them in the project's database settings. - An OpenAI API key. - A database user that can only read. See Good to know. ## Set it up 1. In n8n, create a workflow, open its menu and choose Import from File. Pick `workflow.json`. 2. Add a Postgres credential with the database's host, database, user and password. Select it in `DB Schema`, `Get table definition` and `Run SQL Query`. 3. Add an OpenAI credential and select it in `OpenAI Chat Model`. 4. Open `AI Agent` and read its system message, three short lines that cast it as a database assistant. Add notes about your tables there if you want the agent to know what a column means. 5. Save, press Chat on the canvas and ask, for example, which tables it can see. ## How a run goes 1. `When chat message received` takes your message, and `AI Agent` reads it from `chatInput`. 2. The agent is an OpenAI Functions agent with three tools. 3. `DB Schema` lists the base tables in the `public` schema. 4. `Get table definition` takes a `table_name` from the model. It returns each column's name, type, nullability and default, with the table a foreign key points to. 5. `Run SQL Query` runs the SQL the model writes in its `query` field. Its description tells the agent to use the `->>` operator to read values out of JSON columns. 6. The agent answers in the chat from the rows that come back. ## Change the look - The agent's instructions: the system message in `AI Agent`. - The model: `OpenAI Chat Model`, which sets none in this file. - What the tools may see: the schema filter, `public`, in `DB Schema` and `Get table definition`. - How the agent writes SQL: the tool descriptions in `Run SQL Query` and `Get table definition`. ## Good to know - `Run SQL Query` runs whatever SQL the model writes, including inserts, updates and deletes. Connect with a database user that can only read, or point it at a copy. - `Get table definition` places the table name the model gives straight into its query text. A read-only user limits what a badly worded or hostile message can do. - Agent type: n8n marks the OpenAI Functions mode as deprecated and says it is removed in n8n v3. After that update, replace `AI Agent` with the current AI Agent and reconnect its model and tools. - Models: `OpenAI Chat Model` sets no model, so it takes the node's own default, `gpt-5-mini` in n8n's repository on 12 November 2026. OpenAI's deprecations page, read that day, gives 14 January 2027 as the shutdown of `gpt-4o` and `gpt-4o-mini`, with `gpt-6-luna` as the replacement. If your n8n offers an older default, pick a model in the node. - Only the `public` schema is listed. Tables in other schemas are not found by the two lookup tools. - Query results go to the model, so a large result costs tokens. Ask for counts and totals instead of whole tables. - Your data: the table list, the column definitions, your questions and the rows returned go to OpenAI. - The chat opens only inside n8n's editor, since the trigger's public option is off. ``` ## What is in it 3 files, 11 KB, downloaded as a zip. ```text NOTICE.md 618 B README.md 4 KB workflow.json 6 KB ``` ## Workflow check Checked by Mindket on 13 Nov 2026. Read from the workflow’s file, not from a run. - Steps: 2 steps, starting on a chat message - AI: 1 step calls an AI model: AI Agent - Sign-ins: Signs in to: OpenAI - API keys: No API key is left in the file - Step types: Mindket’s checker reads 1 of its 2 step types. It does not read: AI Agent ## Get it - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/9fb0fe04-9071-4c83-9ccd-8366be2301f8) - [Its page on Mindket](https://mindket.com/p/9fb0fe04-9071-4c83-9ccd-8366be2301f8)