How to Configure Windsurf Redshift Sandbox
To set up a windsurf redshift sandbox, create a restricted database user. Then let the editor use those local keys to run queries. Teams often connect their AI coding agent to a test warehouse. They give the agent a basic login role to read test tables. When that role is too broad, the agent can change shared tables. It can also pull sensitive rows into your local chat window. How a windsurf redshift sandbox works in practice Cognition renamed Windsurf to Devin Desktop in June 2026. The app k
Your coding agents run with your access.
Claude Code reaches prod with a developer's credentials, and approval prompts get clicked through. hoop.dev checks every query, masks sensitive data, and holds risky changes for a human.
To set up a windsurf redshift sandbox, create a restricted database user. Then let the editor use those local keys to run queries. Teams often connect their AI coding agent to a test warehouse. They give the agent a basic login role to read test tables. When that role is too broad, the agent can change shared tables. It can also pull sensitive rows into your local chat window.
How a windsurf redshift sandbox works in practice
Cognition renamed Windsurf to Devin Desktop in June 2026. The app keeps the same editor features and agent workflow. Its built-in agent can run terminal commands with your permission. It opens a dedicated shell on your computer. It uses zsh on macOS and bash on Linux. That shell loads your personal profile, aliases, and local variables.
If your shell profile holds warehouse keys, the agent inherits them. The legacy Cascade agent also links to local Model Context Protocol servers. Both paths let the agent send queries straight to your test cluster. You set up a dedicated database user for this sandbox work. You grant that user read permissions on specific schemas. The agent then uses those local keys to inspect tables and test SQL.
What native sandbox roles solve and what they miss
Native database roles decide who can log in to the cluster. They also control which schemas and tables a user can read. A sandbox role keeps the AI agent out of production tables. It stops the agent from dropping core warehouse schemas by mistake. This is a good first step for any data team.
Still, static database permissions only check access at login and table level. They do not inspect what individual SQL statements do in flight. Once the agent logs in, it can run heavy queries or dump full tables. Test databases often hold real copies of customer emails and phone numbers. They might also hold live card numbers or national IDs from recent snapshots.
Redshift Security + AI Sandbox Environments: architecture patterns and best practices
Continue reading? Get the full guide.
Free. No spam. Unsubscribe anytime.
When the agent queries those tables, raw values flow back to the editor. Static roles also require you to store database keys on developer laptops. Every local shell script and MCP server can read those saved secrets. You cannot pause a single risky query for human review with native roles alone. You also lack a clear audit log that ties each query to a person.
Keeping test data safe during AI agent sessions
Safe sandbox setups separate identity checks from local shell files. Developers should not store long-lived warehouse passwords in their shell profiles. Instead, teams need controls that check every query before it runs. Field-level masking can rewrite emails and card numbers in query results. This keeps sensitive test data out of third-party model prompts. Teams also need clear logs showing every query the agent sends.
Govern statements and responses across databases and APIs
Good config and scoped roles are needed setup. They do not check single queries in flight or mask sensitive fields in returned rows. Across Postgres, MySQL, SQL Server, MongoDB, ClickHouse, Kubernetes APIs, SSH, and internal HTTP endpoints, hoop.dev sits inline in the protocol beside each supported service as an open-source proxy, a sidecar for your database or API. For every query or command an AI agent or engineer runs against production, hoop.dev decides to pass, mask, block, or hold it for human approval.
First, hoop.dev guardrails block banned queries before they run or hold risky ones for review, and hoop.dev data masking rewrites matched sensitive values in responses in memory. Furthermore, hoop.dev holds SOC 2 Type II and is open source on GitHub.
Frequently asked questions
How does the Windsurf agent reach my Redshift test cluster?
The agent runs commands in a local shell or through MCP servers. It uses the database keys and variables stored in your shell profile.
Why is a read-only database role not enough for a sandbox?
A read-only role still lets the agent pull full tables of data. If your sandbox has real emails or phone numbers, they reach the model.
Did Windsurf change its name to Devin Desktop?
Yes, Cognition renamed Windsurf to Devin Desktop in June 2026. The editor keeps the same features, shell access, and Cascade agent tools.
Related guides
In this post