postgresql
PostgreSQL
Paste postgresql:// once — browse schemas, edit rows, map foreign keys, and ask Copilot with schema context.
postgresql://user:password@host:5432/databaseThe story
Maria runs three PostgreSQL databases for her agency — staging, production, and a client warehouse that nobody wants to touch on Friday afternoons. For years she juggled psql, pgAdmin, and a spreadsheet of connection strings that lived in Slack DMs. The breaking point came when a junior dev pasted a production URL into a staging migration script and spent an hour rolling back a harmless-looking ALTER TABLE.
She opened Nexoxa Bridge, pasted postgresql:// once, and gave each database a human name: Agency Staging, Agency Prod, Client Warehouse. The next morning she connected to Client Warehouse before standup. Pulse showed active sessions; the catalog tree loaded every schema object without a single \d command; and before she wrote one JOIN, the knowledge graph painted FK communities across invoices, clients, and line items — god nodes glowing where migrations would hurt most.
That is the Postgres story in Bridge: one vault, one studio, schema context that travels with you instead of living in your head.
Connection URL
Scheme: postgresql://
Parameters
postgresql://user:password@host:5432/database — postgres:// is also accepted — normalized to postgresql:// internally.
Step 1
Schema explorer
Maria's first stop is never the data grid — it is the schema tree. She expands public, then auth, then the client's custom schema where naming conventions went sideways in 2019. Tables, materialized views, RLS policies, extensions, sequences, and enums appear in one hierarchy. She clicks a table and sees column types, defaults, and comments without opening information_schema.
When a client asks whether a column is nullable, she answers in ten seconds instead of writing a four-line SQL probe.
Bridge introspects PostgreSQL catalog depth on connect: tables, views, materialized views, indexes, foreign keys, sequences, enums, composite types, functions, triggers, extensions, and row-level security policies. The tree mirrors what you would assemble from pg_catalog queries, but stays live beside your query tabs. Saved connection names encrypt in Windows Credential Manager — MCP and Copilot reference the name, never the raw URL with embedded passwords.
Step 2
Data grid & SQL lab
A support ticket claims invoice totals are wrong for Q2. Maria filters the invoices table to April–June, sorts by updated_at, and spots three rows with null client_id — orphans that should have been caught by a FK. She edits inline, follows the foreign key link to clients, and confirms the parent rows exist.
She switches to Query lab, runs EXPLAIN on her fix query, and shares the plan screenshot in Slack. No CSV export round trip, no second tool.
The data grid supports type-aware inline editing, bulk filters, and FK navigation to parent tables in one click. CSV import and export live in the same workspace. Query lab provides syntax highlighting, result grids, and EXPLAIN plans. Switch between saved Postgres connections from the Connections grid without reopening the app.
Step 3
Knowledge graph
Before a risky migration on Client Warehouse, Maria opens the knowledge graph. FK edges cluster invoices with clients and products; a god node at clients tells her half the schema hangs off that table. She screenshots graph.html for the change request and schedules the migration for Tuesday instead of Friday.
The graph did not phone home — it built from metadata on her laptop the moment she connected.
The database knowledge graph renders interactively in graph.html: FK communities, hub tables (god nodes), and column indexes for AI context. Reports land in db-knowledge-out/ for Copilot and Cursor rules. Credentials never leave your machine; only schema metadata feeds the graph.
Step 4
Copilot with schema context
Maria asks Copilot in plain English: which tables reference clients.id and what ON DELETE rules apply? The answer cites foreign-keys from the catalog and suggests a safe migration order. She runs the suggested SQL in Query lab, tweaks one line, and ships.
Bring-your-own-key means her OpenAI usage stays on her account — schema context is local, not a shared cloud index.
AI Copilot grounds answers in the active connection's schema graph and catalog. Ask for JOINs, index advice, or migration steps; open results in Query lab. BYOK configuration keeps API keys in your control.
Next
Explore the full feature guide: PostgreSQL guide