dryrun

The Offline-First PostgreSQL MCP That Won't Ruin Your Production

Schema linting, migration safety checks, query validation, and 16 MCP tools for your AI assistant, all working from a JSON snapshot. No production credentials needed.

Get dryrun GitHub Repository

How It Works

# Dump schema from production (once)
dryrun dump-schema --source "$DATABASE_URL" -o schema.json

# Initialize project and import snapshot
dryrun init
dryrun import schema.json

# Lint naming conventions, PK types, missing indexes...
dryrun lint

# Detect drift between saved snapshot and live database
dryrun --db "$DATABASE_URL" drift

# Serve as MCP server for Claude or other AI assistants
dryrun mcp-serve

Features

Schema Linting

19 convention rules covering naming, primary key types, varchar vs text, missing timestamps, FK indexes, and partition health. 14 structural audit rules for duplicate indexes, circular FKs, bloat detection, and more. Configurable severity and rule disabling.

Migration Safety

Analyzes DDL statements for lock types, table rewrite risks, and index build impact. Version-aware behavior, safe alternatives, and rollback DDL for every dangerous operation.

Query Validation

Parses SQL via libpg_query against your actual schema. Catches non-existent columns, detects anti-patterns, and suggests missing indexes, all without hitting the database.

Multi-Node Stats

Collect per-replica statistics, detect seq_scan hotspots and routing imbalances across your cluster. Aggregate stats from primaries and standbys into a single snapshot.

MCP Server for AI

16 tools that give Claude and other AI assistants full schema awareness. Schema exploration, query validation, migration checks, vacuum health, drift detection, and more. Supports stdio and SSE transports.

Offline-First

Dump the schema once, commit the JSON, and everyone works offline. No production credentials on developer machines or CI runners. Optional live DB connection for EXPLAIN and drift detection.

Installation

Homebrew

brew install boringsql/boringsql/dryrun

Cargo (Rust)

cargo install dryrun --git https://github.com/boringsql/dryrun

Binary releases coming soon.

Why dryrun?

AI coding assistants write SQL blind. They don't know your schema, your indexes, or your constraints, so they generate migrations that take ACCESS EXCLUSIVE locks on busy tables, add varchar(255) columns when you use text, or create indexes that already exist.

dryrun makes your database schema portable:

Capture once, use everywhere. One dump-schema command produces a JSON file that contains your tables, columns, indexes, constraints, functions, enums, RLS policies, triggers, extensions, and per-table statistics. Commit it, share it, CI it.
Lint before you ship. 33 rules (19 convention + 14 audit) catch naming violations, missing timestamps, serial vs identity, duplicate indexes, circular FKs, and more, all from the JSON file, no database connection needed.
Check before you migrate. Every DDL statement gets a lock type, rewrite risk, safe alternative, and rollback DDL. Version-aware: knows which operations are safe on your PostgreSQL version.

Use Cases

AI-assisted database development

Install dryrun as an MCP server and your AI assistant gains full schema awareness. It stops hallucinating column names, suggests proper indexes, and warns you when a migration will rewrite the table.

Pre-deploy migration review

The check_migration MCP tool analyzes every DDL statement for lock types and rewrite risks. Wire it into your AI-assisted workflow or CI pipeline.

Schema governance across teams

Enforce naming conventions, require timestamps, ban varchar in favor of text, and flag tables without primary keys. One dryrun.toml config keeps the whole team aligned.