Quickstart

Two ways to use rawquery. Pick one.

• Path A. CLI + coding agent (recommended). You describe what you want in plain English. Claude Code or Codex drives the CLI for you: connects sources, writes SQL, creates charts, publishes pages.
• Path B. Manual. You drive it yourself. Same capabilities, no AI layer. Works from the terminal (rq) or from the dashboard.

Everything below applies to both paths. The flow is the same.

Path A. CLI + coding agent

Three commands, then you talk to the agent.

1. Install and sign up

curl -sSL rawquery.dev/install.sh | sh
rq signup

curl -sSL rawquery.dev/install.sh | sh
rq signup

Single binary, zero dependencies. rq signup creates the account, the workspace, and the API key in one call. Already have an account? rq login.

2. Install the agent skill

For Claude Code:

rq claude-init

rq claude-init

For Codex:

rq codex-init

rq codex-init

claude-init drops .claude/skills/rq/SKILL.md; codex-init drops .agents/skills/rq/SKILL.md plus agents/openai.yaml. Either way the agent gets a live view of your workspace: schemas, tables, connections, saved queries, transforms, charts, pages.

3. Ask

claude   # or: codex

claude   # or: codex

Trigger the skill explicitly: /rq in Claude Code, $rq in Codex. Or skip the prefix — both agents will pick the skill up implicitly when you mention rawquery, your data, or SQL. Then:

> connect my Stripe account, here is the key sk_live_xxx
> what tables do I have?
> show me top 10 customers by revenue
> build me a dashboard with monthly revenue, churn, and signups
> publish it

The agent explores your schema, writes the SQL, runs it, saves queries, creates charts, assembles pages, publishes them. You get the URL back.

The same skill works in Cursor and Windsurf. If you cannot install the CLI (remote setup, Cowork, etc.), connect to the hosted MCP server instead. See Claude Code & Codex and Claude (web).

Path B. Manual

Same result, you drive.

Option 1. CLI

curl -sSL rawquery.dev/install.sh | sh
rq signup
rq connect stripe --token sk_live_xxx --schedule daily
rq tables
rq query "SELECT * FROM my_stripe.customers LIMIT 10"

curl -sSL rawquery.dev/install.sh | sh
rq signup
rq connect stripe --token sk_live_xxx --schedule daily
rq tables
rq query "SELECT * FROM my_stripe.customers LIMIT 10"

Full command list in the CLI reference.

Option 2. Web UI

1. Sign up. Free, no credit card.
2. Connections, Add Connection. PostgreSQL, MySQL, Stripe, HubSpot, Salesforce, Shopify, Google Sheets, or any HTTP/GraphQL API via custom spec. Credentials are encrypted at rest.
3. Wait 1 to 5 minutes for the first sync. Watch progress on the Connections page.
4. Open the Lakehouse to browse tables, previews, and column types.
5. Open the Query editor. Monaco-powered, cost estimation before you run, CSV and JSON export after.

Tables are namespaced by connection name: my_stripe.charges, crm.contacts.

SELECT
  customer_email,
  COUNT(*) as order_count,
  SUM(amount) as total_spent
FROM my_stripe.charges
WHERE status = 'succeeded'
GROUP BY customer_email
ORDER BY total_spent DESC
LIMIT 10

SELECT
  customer_email,
  COUNT(*) as order_count,
  SUM(amount) as total_spent
FROM my_stripe.charges
WHERE status = 'succeeded'
GROUP BY customer_email
ORDER BY total_spent DESC
LIMIT 10

See it run

Next

• Transforms. SQL models on top of raw data. Scheduled, DAG-aware.
• Saved queries as API. Parameterized SQL in, JSON out via curl.
• Charts and pages. Shareable URLs, password-protect, auto-refresh.
• Execute API. curl -X POST https://api.rawquery.dev/api/v1/execute with an API key.
• CLI reference. Every rq command and flag.

Need help? hello@rawquery.dev. Answered within hours.