planetscale-cli-skills

PlanetScale CLI command reference and workflows. Use for authentication; org, SSO, teams, and billing; databases, branches, Neki, internal or external keyspaces, keyspace parameters, logs, maintenance, extensions, metrics, insights, inspect, and SQL; deploy requests, schema migrations, MoveTables, throttlers, rollout concurrency, disk-storage settings, and Traffic Control; PgBouncers, dedicated read replicas, roles, passwords, backups, webhooks, audit logs, service tokens, D1 imports, and automation. Routes to focused pscale sub-skills. Triggers on PlanetScale CLI, pscale, pscale --skill, database, branch, Neki, Neki read-only replicas, --read-only-replicas, keyspace, keyspace parameters, create-external, external keyspace, deploy request, move-tables, vtctl, vtctld, max rollout, disk scaling, max storage, throttler, traffic-control, metrics, insights, inspect, SQL, role, password, pgbouncer, dedicated read replica, read-only replica, backup, webhook, audit-log, billing, org, SSO, service token, import d1.

Comprehensive pscale command reference and workflows for managing PlanetScale databases via terminal.

Overview

The PlanetScale CLI brings database branches, deploy requests, and schema migrations to your fingertips. This skill provides command references, automation scripts, and decision trees for all pscale operations.

Configuration and credential safety

• Directory-local .pscale.yml files are project configuration, not trusted credential configuration. The CLI accepts only org, database, and branch from them and warns while ignoring keys such as API endpoints or tokens.
• Keep API URLs and credentials in the user config, environment/secret manager, or explicit approved flags. Never commit them to a repository-local config.
• pscale api follows cross-host redirects without forwarding authentication or caller-supplied headers. Even with this protection, pass secret-bearing headers only to an explicitly verified API host and do not expose them in logs.
• When --format json is active, raw API error codes are preserved in issues[0].code instead of being collapsed to COMMAND_FAILED. Branch automation should branch on that exact nested code when present.
• schema_mutation_blocked, schema_refreshing, and schema_snapshot_not_ready responses use status: "action_required" and one CLI-generated next_steps instruction. Treat them as retryable waits, not hard failures or permission to retry blindly: wait for the active vtctld mutation or deploy to finish, or for the branch schema snapshot to become ready, then re-read state before retrying.
• A Postgres resize in progress error also yields status: "action_required" with wait/status guidance. Inspect pscale branch resize status, wait for the existing change to finish, and re-read the intended state before retrying; do not re-authenticate or cancel another operator's change as a shortcut.
• Route MySQL/Vitess schema snapshot refresh/readiness to pscale-branch: refresh-schema waits for schema_ready with 2-second polling and a 2-minute readiness bound. On schema_refreshing or schema_snapshot_not_ready, read back readiness before retrying the follow-up reader; preserve poll failures rather than treating every error as a snapshot timeout.
• Route single-request create/update --aggressive-cutover to pscale-deploy-request. It can kill blockers on the first cutover attempt and requires explicit approval plus read-back; database defaults and emergency force-cutover are separate actions.
• Route parameter update-requirement interpretation to pscale-branch for Postgres catalogs and pscale-neki for configuration-profile, sidecar, and admin catalogs. REQUIRES UPDATE is a human boolean indicator; JSON requires_update preserves a nullable string. Inspect non-null requirements before proposing writes; they do not authorize automatic maintenance or capacity changes.
• Route Postgres branch extension enable/disable and Vitess VTGate parameter/update/change-rollout work to pscale-branch. Route approved failed deploy-operation retries to pscale-deploy-request; retry is not queue unblocking or revert. Keyspace storage size units and staff-managed maximum limits belong to pscale-database.
• Route Neki read-only replica capacity to pscale-neki and branch config-profile create/update --read-only-replicas. These per-shard counts are not dedicated PostgreSQL replica resources; confirm the engine and retain exact-target capacity/cost approval.
• Route Neki live router sessions to pscale-branch (branch connections show/top): read-only, no kill actions or engine-incompatible target/filter flags; preserve incomplete-result warnings. Sidecar settings belong to pscale-neki and config-profile update --parameters sidecar.<name>=<value>; branch sidecar is deprecated. Parameter inventories expose INTERNAL / JSON internal; internal settings are not write candidates, and false alone is not authorization.

Binary-native agent bootstrap

• pscale --skill prints the current binary's installable pscale-cli skill file to stdout when no positional arguments follow it. pscale agent-guide --skill is equivalent.
• Both forms still emit the raw Markdown skill when --format json is present; redirect the output to an approved skills directory when a standalone binary-native guide is wanted.
• Validate automated bootstrap output starts with ---\nname: pscale-cli; do not infer a valid skill document from exit status alone.
• The binary-native guide is useful for first-party command discovery, but it does not replace this repository's specialized safety workflows, focused references, and helper scripts.
• Unrecognized root tokens and subcommands of non-runnable command groups fail with exit status 2 instead of printing help and exiting zero. Runnable groups such as keyspace parameters can interpret an unrecognized token as a positional argument and execute their default action, so automation must validate the expected JSON/resource shape as well as the exit code.
• The top-level pscale data-imports family is unavailable. Use pscale keyspace create-external to attach an external MySQL source and pscale branch vtctl move-tables for table-movement workflows; treat an unknown-command error as a failed operation.

Sub-Skills

Command	Skill	Use When
auth	pscale-auth	Login, logout, service tokens, authentication management
branch/logs	pscale-branch	Create, restore to a PostgreSQL recovery point, rename, protect, delete, promote, diff, list, and switchover branches; refresh MySQL schema snapshots and verify readiness; query PostgreSQL/Neki branch logs; inspect branch infra, manage Postgres size/replicas/parameters/maintenance/extensions, toggle Postgres extensions, run Postgres/Neki branch maintenance, resize Vitess VTGates and roll out VTGate parameter changes, manage live keyspace routing rules, Lookup Vindexes, and tablet throttling, manage query pattern reports, manage Vitess MoveTables workflows
deploy-request	pscale-deploy-request	Create, review, inspect queues/operations, check storage, throttle, deploy, update auto-apply/auto-delete/per-request aggressive-cutover settings, retry failed operations after diagnosis, unblock failed deploy/revert queues, force cutover, and revert schema changes
database	pscale-database	Create, list, show, update, delete, and dump databases; manage database-level and keyspace-level Vitess throttlers, keyspace VTTablet/MySQL parameters, rollout concurrency, disk-storage settings, internal/external keyspaces, aggressive-cutover defaults, PostgreSQL IP restrictions, and read-only regions
neki	pscale-neki	Manage Neki data topology, shards, branch-wide change requests, config profiles, per-shard read-only replica capacity (--read-only-replicas) and extension sets, routers, sidecars, admin config, profile maintenance, restore sizing overrides, target-specific inspection, and router/shard role access
maintenance	pscale-maintenance	Inspect Vitess Enterprise maintenance schedules, pending versions, deadlines, and historical windows
sql	pscale-sql	Run non-interactive SQL queries with JSON output and ephemeral credentials
metrics	pscale-metrics	Query historical/current branch metrics and engine-aware grouped performance reports
insights	pscale-insights	Analyze production query statistics, execution samples, query tags, error details, anomaly correlations, and schema recommendations
traffic-control	pscale-traffic-control	Inventory and safely manage Postgres Traffic Control budgets and fingerprint/keyspace/tag rules
inspect	pscale-inspect	Run point-in-time, read-only MySQL/Vitess, PostgreSQL, and Neki diagnostic checks against one connection target
import d1	pscale-import-d1	Import Cloudflare D1 SQLite exports into PlanetScale Postgres
backup	pscale-backup	Create, list, show, restore, and delete branch backups; manage scheduled backup policies
billing	pscale-billing	Inspect invoices and manage organization payment methods with sensitive billing-data safeguards
audit-log	pscale-audit-log	List audit events and export filtered authentication attempts
password	pscale-password	Create, list, show, update, delete, and scope Vitess connection passwords to read-only regions; inspect and mutate Postgres/Neki roles, inherited-role pairings, and connection targets
pgbouncer	pscale-pgbouncer	List, inspect, create, resize, cancel, and delete dedicated PostgreSQL PgBouncers
dedicated-read-replica	pscale-read-only-replica	List, inspect, create, resize, configure, and delete dedicated PostgreSQL read replicas; recognize the deprecated read-only-replica compatibility name
webhook	pscale-webhook	List, inspect, create, update, test, and delete database webhooks, including authorization headers
org	pscale-org	List, show, switch, and update organizations; manage SSO, directory sync, email domains, members, and teams
service-token	pscale-service-token	Create, show, and manage CI/CD service tokens

Decision Trees

Should I use a branch or deploy request?

text

What's your goal?
├─ Experimenting with schema changes → Create branch (pscale-branch)
├─ Testing schema in isolation → Create branch (pscale-branch)
├─ Ready to deploy schema to production → Create deploy request (pscale-deploy-request)
└─ Reviewing schema changes before production → Review deploy request (pscale-deploy-request)

Service token vs password?

text

What's your use case?
├─ CI/CD pipeline → Service token (pscale-service-token)
├─ Local development → Password (pscale-password)
├─ Production application → Service token (rotatable, secure)
└─ One-off admin task → Password (temporary)

Direct promotion vs deploy request?

text

Production readiness?
├─ Immediate promotion (dangerous) → pscale branch promote (pscale-branch)
├─ Review + approval workflow → pscale deploy-request create (pscale-deploy-request)
└─ Safe production deployment → Always use deploy requests

Common Workflows

Schema Migration Workflow

Complete workflow from branch creation to production deployment:

bash

# 1. Create development branch
pscale branch create <database> <branch-name>

# 2. Make schema changes (via shell, ORM, or direct SQL)
pscale shell <database> <branch-name>

# 3. View schema diff
pscale branch diff <database> <branch-name>

# 4. Create deploy request
pscale deploy-request create <database> <branch-name>

# 5. Review and deploy
pscale deploy-request deploy <database> <deploy-request-number>

# 6. Verify deployment
pscale deploy-request show <database> <deploy-request-number>

See scripts/ directory for automation.

Branch Development Workflow

bash

# Create branch from main
pscale branch create <database> <feature-branch> --from main

# Work on schema changes
pscale shell <database> <feature-branch>

# Check diff before deploying
pscale branch diff <database> <feature-branch>

# Create deploy request when ready
pscale deploy-request create <database> <feature-branch>

CI/CD Integration

bash

# Create service token for CI/CD
pscale service-token create --org <org>

# Use in CI/CD pipelines (GitHub Actions, GitLab CI, etc.)
export PLANETSCALE_SERVICE_TOKEN_ID=<token-id>
export PLANETSCALE_SERVICE_TOKEN=<token>

# Create and deploy via CI/CD after review/approval gates pass
pscale deploy-request create <database> <branch> --format json
pscale deploy-request deploy <database> <deploy-request-number>

Cloudflare D1 to PlanetScale Postgres import

bash

# Lint and dry-run first; review the JSON migration ID and warnings
pscale import d1 lint --input ./d1-export.sql --format json
pscale import d1 start <database> <branch> --input ./d1-export.sql --dry-run --format json

# After explicit confirmation, run and verify the import
pscale import d1 start <database> <branch> --input ./d1-export.sql --migration-id <id> --format json
pscale import d1 verify <database> <branch> --migration-id <id> --input ./d1-export.sql --format json

Quick Reference

Most Common Commands

bash

# Authentication
pscale auth login
pscale auth logout

# Branch management
pscale branch create <database> <branch> [--from <source-branch>]
pscale branch list <database>
pscale branch delete <database> <branch>
pscale branch parameters list <database> <branch> --format json
pscale branch resize status <database> <branch> --format json
pscale branch vtgate show <database> <branch> --format json
pscale branch update <database> <branch> --deletion-protected=true --format json

# Discover and use a Vitess read-only region
pscale keyspace read-only-regions <database> <branch> <keyspace> --format json
pscale keyspace read-only-regions add <database> <branch> <keyspace> <region> --cluster-size <size> --replicas <count>
pscale password create <database> <branch> <name> --read-only-region <region> --format json

# Deploy requests
pscale deploy-request create <database> <branch>
pscale deploy-request list <database>
pscale deploy-request storage-check <database> <number> --format json
pscale deploy-request operations <database> <number> --format json
pscale deploy-request deploy <database> <number>

# Dedicated PostgreSQL PgBouncers
pscale pgbouncer list <database> <branch> --format json
pscale pgbouncer show <database> <branch> <name> --format json

# Dedicated PostgreSQL read replicas
pscale dedicated-read-replica list <database> <branch> --org <org> --format json
pscale dedicated-read-replica show <database> <branch> <name> --org <org> --format json

# Database operations
pscale database create <database> --org <org>
pscale database list
pscale database show <database> --format json
pscale database ip-restriction list <database> --format json
pscale database throttler show <database> --org <org> --format json
pscale database aggressive-cutover show <database> --org <org> --format json
pscale keyspace settings <database> <branch> <keyspace> --org <org> --format json
pscale keyspace parameters list <database> <branch> <keyspace> --org <org> --format json
pscale shell <database> <branch>
pscale logs <database> <branch> --org <org> --format json

# Database webhooks
# Webhook resource JSON may include the signing secret; redact before capture.
set -o pipefail
pscale webhook list <database> --org <org> --format json |
  jq 'map(del(.secret))'
# Use `show` only when intentionally retrieving the secret; otherwise use the redacted sub-skill workflow.

# Maintenance schedules and windows (Vitess Enterprise)
pscale maintenance list <database> --org <org> --format json
pscale maintenance windows <database> <schedule-id> --org <org> --format json

# Backup policies and authentication-attempt exports
pscale backup policy list <database> --format json
pscale audit-log auth-attempts download --org <org> --since 24h --outcome deny

# Non-interactive read query for agents/scripts
pscale sql <database> <branch> --org <org> --format json --query "SELECT 1"

# Point-in-time diagnostics, branch metrics, and query-fingerprint analysis
pscale inspect all <database> <branch> --org <org> --format json
# Neki: pin one shard on the primary path by default
pscale inspect all <database> <branch> --org <org> --shard <shard-id> --format json
# Add --router <router> and/or --replica only for that explicit target path
pscale metrics report <database> <branch> --org <org> --period 1d --format json
pscale metrics show <database> <branch> --org <org> --metric queries --metric latency_p99 --period 1h --format json
pscale insights queries <database> <branch> --org <org> --sort p99Latency --period 1h --format json
pscale insights queries samples <database> <branch> <fingerprint> --org <org> --keyspace <keyspace> --format json
pscale insights tags summaries <database> <branch> --org <org> --tags app --sort totalTime --format json
pscale insights errors show <database> <branch> <full-error-fingerprint> --org <org> --format json
pscale insights anomalies show <database> <branch> <anomaly-id> --org <org> --format json
pscale insights recommendations <database> --org <org> --format json
pscale insights recommendations show <database> <number> --org <org> --format json
pscale insights queries show <database> <branch> <query-id> --org <org> --format json
pscale insights queries summary <database> <branch> <fingerprint> --org <org> --keyspace <keyspace> --format json

# Postgres Traffic Control inventory
pscale traffic-control budget list <database> <branch> --org <org> --format json

# Organization billing inspection
pscale billing invoice list --org <org> --page 1 --per-page 25 --format json
pscale billing payment-method show --org <org> --format json

# Cloudflare D1 import dry-run
pscale import d1 start <database> <branch> --input ./d1-export.sql --dry-run --format json

Related Skills

• drizzle-kit - ORM schema management and migrations
• gitlab-cli-skills - GitLab MR workflow integration
• github - GitHub PR and CI/CD integration

Automation Scripts

See scripts/ directory for token-efficient automation:

• create-branch-for-mr.sh - Create PlanetScale branch matching your MR/PR branch name
• deploy-schema-change.sh - Complete schema migration workflow
• sync-branch-with-main.sh - Create a replacement branch from main/base for conflict resolution

Scripts execute without loading into context (~90% token savings).

Resources

• Official docs: https://planetscale.com/docs/reference/planetscale-cli
• GitHub: https://github.com/planetscale/cli
• Community: https://github.com/planetscale/discussion