SQL Server CLI for AI coding agents.

One install. Your agents automatically know how to inspect SQL Server databases, run SQL, and export results.

sql is the canonical raw-SQL surface in this project.

You can run raw SQL either with the explicit subcommand:

sscli sql "SELECT 1"

or with the top-level shorthand:

sscli "SELECT 1"

For configuration, prefer explicit --env-file over relying on ambient .env files in the working directory.

sqlcmd is a great general-purpose SQL Server client, especially for interactive sessions and ad-hoc work.

For tool-calling agents, sqlcmd tends to be a poor fit because it's optimized for humans, not for structured, repeatable automation:

• Output is hard to consume: sqlcmd output is human-oriented text; agents usually want stable markdown tables or a single JSON object they can reliably parse.
• Schema discovery is manual: you end up writing catalog queries (sys.tables, INFORMATION_SCHEMA, etc.) instead of calling purpose-built primitives like sscli tables, sscli describe, and sscli columns.
• Structured SQL workflows: sscli keeps raw SQL, schema discovery, and machine-readable output in one tool, so agents do not have to switch between sqlcmd for execution and a second tool for inspection.
• More setup friction: sqlcmd is typically installed via Microsoft tooling and may require ODBC drivers depending on platform/CI image; sscli is a single binary with config + env var discovery built in.
• No agent integration: sscli can install a reusable skill/extension so agents "know the tool" without you pasting usage docs into every prompt.

Keep sqlcmd for interactive SQL. Reach for sscli when you want raw SQL plus fast schema inspection and output formats that are easy for agents to use.

1. Install sscli

# macOS/Linux
brew install jwcraig/tap/sscli

# Windows (PowerShell)
scoop bucket add jwcraig https://github.com/jwcraig/scoop-bucket
scoop install sscli

# or with cargo (any platform)
cargo install sscli

2. Teach your agents

sscli integrations skills add --global   # Claude Code + Codex
sscli integrations gemini add --global   # Gemini CLI

Done. Your agents now know how to browse schemas, run SQL, and export results.

For humans who want to use sscli directly.

# Create a starter config in ./.sql-server/config.yaml (safe defaults)
sscli init

# Set the password env var referenced by passwordEnv in your config. sscli also reads
export SQL_PASSWORD='...'

# Sanity-check connectivity + server metadata
sscli status

# See the effective settings + which config file was used
sscli config

sscli status                              # Check connectivity
sscli tables                              # List tables
sscli tables --like "%User%" --describe   # Describe all User-related tables
sscli describe Users                      # DDL, columns, indexes, triggers
sscli describe T_Users_Trig               # Trigger definition (auto-detected)
sscli table-data equipment                # Browse rows (schema auto-resolved; prompts on conflicts)
sscli sql "SELECT TOP 5 * FROM Users"
sscli "SELECT COUNT(*) FROM Users"        # Top-level shorthand for inline SQL
sscli sql --file [path/to/file]           # Run long queries, execute bulk statements
cat patch.sql | sscli sql --stdin         # Pipe a script on stdin
sscli update                              # Check for new releases (alias: sscli upgrade)

brew install jwcraig/tap/sscli

scoop bucket add jwcraig https://github.com/jwcraig/scoop-bucket
scoop install sscli

curl -sSL https://raw.githubusercontent.com/jwcraig/sql-server-cli/main/install.sh | sh

The installer verifies the downloaded artifact against the release checksums-sha256.txt.

cargo binstall sscli

cargo install sscli

Download from GitHub Releases.

git clone https://github.com/jwcraig/sql-server-cli sscli
cd sscli
cargo build --release
./target/release/sscli --help

# Check if you're up to date (alias: `sscli upgrade`)
sscli update

# Homebrew
brew upgrade sscli

# Cargo
cargo install sscli --force

By default, sscli does not check for updates automatically.

To enable lightweight update notifications (stderr, TTY-only, cached), create:

• ~/.config/sscli/settings.json (Linux/XDG default)
• macOS often uses ~/Library/Application Support/sscli/settings.json by default

Example settings.json:

{ "autoUpdate": true }

The installed skill file tells agents:

• When to use sscli (database inspection, schema discovery, raw SQL execution)
• Available commands and their purpose
• Output preferences (markdown for context efficiency, --json for structured data)
• sql as the main raw-SQL surface, with top-level shorthand for simple inline queries

sscli supports three ways to configure a connection (highest priority wins; env vars are skipped if you pass --profile):

# 1) CLI flags (one-off / scripts)
sscli status --server localhost --database master --user sa --password '...' # alias: --host

# 2) Environment variables (CI-friendly)
export SQL_SERVER=localhost SQL_DATABASE=master SQL_USER=sa SQL_PASSWORD='...'
sscli status

# 3) Config file (recommended for repeated use)
sscli init && export SQL_PASSWORD='...' && sscli status

Generate a commented template (writes ./.sql-server/config.yaml by default):

sscli init

Or copy the example file in this repo:

mkdir -p .sql-server
cp config.example.yaml .sql-server/config.yaml

1. --config <PATH>
2. SQL_SERVER_CONFIG / SQLSERVER_CONFIG
3. Walk up from CWD looking for .sql-server/config.{yaml,yml,json} or .sqlserver/config.{yaml,yml,json}
4. Global config: $XDG_CONFIG_HOME/sql-server/config.{yaml,yml,json} (platform-dependent)
5. Environment variables (only applied when no --profile is provided)
6. Hardcoded defaults

Run sscli config to confirm which config file is being used and what values are in effect.

defaultProfile: default
profiles:
  default:
    server: localhost
    port: 1433
    database: master
    user: sa
    passwordEnv: SQL_PASSWORD
    encrypt: true
    trustCert: true

For a fully commented example (including settings.output.*, timeout, and defaultSchemas), see config.example.yaml.

Environment variables override values from the config file when no explicit --profile was passed. If you pass --profile <name>, the profile values win over env vars (flags still win over both).

.env file support: use --env-file to load environment variables from a specific file, for example --env-file .env.dev. sscli does not implicitly load .env from the current working directory.

sqlcmd compatibility: The following sqlcmd environment variables are also supported:

Core (shown in --help):

Advanced (shown in help --all):

Note: sscli sessions filters by client host name using --client-host. --host is reserved as an alias for --server.

JSON output emits exactly one object to stdout. Errors go to stderr.

• keep sql as the canonical raw-SQL command
• support full SQL execution, including file/stdin-driven scripts
• keep the existing profile/config connection model
• prefer explicit --env-file over ambient cwd configuration
• prefer lightweight safety rails and explicit target visibility over hidden parser restrictions

If you need a locked-down distribution, maintain a custom build or wrapper that strips write capability. The shipped tool is intended to be full-capability.

Each command returns a stable top-level object:

Errors (stderr):

{ "error": { "message": "...", "kind": "Config|Connection|Query|Internal" } }

Detects drift between two profiles or explicit connection strings.

Synopsis:

sscli compare --target <profile> [--source <profile>] [--schema web --schema dbo] \
  [--summary|--json] [--ignore-whitespace] [--strip-comments] \
  [--object dbo.ProcName] [--apply-script [path|-]] [--include-drops]

• --target/--right (required): profile to treat as the environment you want to align.
• --source/--left: reference profile (defaults to global --profile or config default).
• --source-connection/--left-connection, --target-connection/--right-connection: override profile with a connection string (URL or ADO-style Server=...;Database=...).
• --schema/--schemas: limit to specific schemas (repeatable or comma-separated).
• --object: emit unified diff for a single module (proc/view/function/trigger).
• --ignore-whitespace, --strip-comments: normalize noise before diffing definitions.
• --summary: compact drift counts; --pretty renders text; --json renders JSON.
• --apply-script [path|-]: generate SQL to align target to source; default path db-apply-diff-YYYYMMDD-HHMMSS.sql in cwd; use - for stdout.
• --include-drops: include DROP statements (disabled by default).
• Profiles are the names in your .sql-server/config.* (e.g., dev, stage, prod). --source/--target expect those names.

Examples:

# Summary with profile names
sscli compare --target prod --summary

# Object diff ignoring whitespace
sscli compare --target prod --object dbo.MyProc --ignore-whitespace

# Apply script to stdout
sscli compare --target prod --apply-script - --include-drops

# Using explicit connection strings instead of profiles
sscli compare --source-connection "Server=dev,1433;Database=app;User ID=sa;Password=..." \
              --target-connection "sqlserver://user:pass@prod:1433/app" \
              --summary

Exit codes: 0 = no drift, 3 = drift detected (summary/object/apply modes), 1 = error.

cargo test

This repo ships a local pre-push hook that runs cargo fmt --check, cargo clippy -D warnings, and cargo test. It’s already enabled via core.hooksPath=.githooks. If you need to bypass temporarily:

HUSKY=0 git push   # or
SKIP=1 git push    # (any env; git ignores but hook can read if we add later)

DB-backed integration tests (opt-in):

SSCLI_INTEGRATION_TESTS=1 SQL_SERVER_CONFIG=/path/to/config.yaml \
SQL_PASSWORD=... cargo test