surrealdb/agent-skills/surrealdb-cli

SurrealDB CLI

A skill for driving the `surreal` command-line binary — SurrealDB's all-in-one tool for running a server, querying it, and moving data in and out.

This skill covers operating the surreal binary itself. To write the SurrealQL that you run with it, use the surrealql skill; for declarative schema migrations and type generation, use the surrealkit skill.

When to use this skill

Reference these guidelines when:

• Starting a SurrealDB server, locally or in a container (surreal start)
• Opening the SQL shell or piping queries from a file (surreal sql)
• Importing or exporting a database for backups, seeding, or migration (surreal import / surreal export)
• Gating scripts or CI on server availability (surreal is-ready)
• Upgrading the binary or repairing on-disk storage after an upgrade (surreal upgrade / surreal fix)
• Importing or exporting SurrealML models (surreal ml)

Installation

Verify with surreal version. Pin a specific image tag (not latest) for production. See the install docs for the current set of methods.

Command map

surreal validate (SurrealQL file checking) and formatting are covered by the surrealql skill. Newer subcommands (mcp, module) also exist — run surreal help and surreal <command> --help to confirm the surface of an installed version.

Connection flags & env vars

sql, import, export, and ml share the same connection flags. Each flag has a SURREAL_* environment variable fallback; an explicit flag wins.

Pass secrets via env vars rather than flags so they do not leak into shell history or process listings.

Quick start

# 1. Start an authenticated in-memory server (foreground)
surreal start --user root --pass root memory
 
# 2. In another shell, open the SQL REPL on namespace/database "main"
surreal sql --endpoint ws://localhost:8000 \
  --user root --pass root --ns main --db main --pretty
 
# 3. Load a dump into that database
surreal import --endpoint http://localhost:8000 \
  --user root --pass root --ns main --db main backup.surql

Rules & conventions

• Authentication is on by default. surreal start needs --user/--pass; only use --unauthenticated for throwaway local instances.
• Choose a storage backend deliberately. memory (the default) is ephemeral — data vanishes on exit. Use rocksdb:/surrealkv:// for persistence and tikv:// for distributed clusters. See start.md.
• Export defaults to stdout. surreal export writes to - (stdout) unless you give a file path; redirect or pipe accordingly.
• Gate CI with `is-ready`. It returns a non-zero exit code until the server accepts connections — ideal in healthchecks and scripts.
• Run `fix` after a major upgrade if the server reports an incompatible storage version, pointing it at the same PATH used by start.
• The CLI evolves. Confirm flags and subcommands against surreal <command> --help for the installed version; this skill reflects v3.x.

References

• Running a server and storage backends — references/start.md
• Interactive and piped querying — references/sql.md
• Importing and exporting data — references/import-export.md
• Versioning, upgrades, readiness, repair, and ML models — references/maintenance.md