Load Database

Instructions

When this skill is invoked — either via /load-db or through a natural language request like "clone the live database from december 12" — download a database backup from Pantheon and load it into the local development environment.

Usage

/load-db                           # Interactive: choose environment and backup
/load-db live                      # Use latest backup from live environment
/load-db dev 2026-02-10            # Use backup from dev matching the given date
"clone the live database"          # Natural language: infer environment=live, pick latest
"load the db from test dec 12"     # Natural language: infer environment=test, date=dec 12
"pull down the production database"# Natural language: default to live, pick latest

Interpreting Natural Language

When invoked through conversation rather than a slash command:

Parse Arguments and Context:
- From slash command: first argument is environment, second is date
- From natural language: extract environment and date from the user's message as described above
- If both environment and date/latest are known, skip interactive selection steps
Select Environment:
- If an environment was already determined, use it directly — skip interactive selection
- Otherwise, list available environments using terminus:
```
terminus env:list ${TERMINUS_SITE} --format=list --fields=id
```
- Use the AskUserQuestion tool to let the user pick an environment from the list
- Present the most commonly used environments first: live, test, dev, qa, then others
- Include up to 3 options in the question, plus the user can type "Other" to specify a different environment
- Recommend live as the default choice since it has production data
Download Backup:
- Latest backup (no date specified): Skip the backup:list step entirely. Use terminus backup:get without --file to get the latest database backup URL directly, then download it in one command:
```
curl -L "$(terminus backup:get ${TERMINUS_SITE}.{environment} --element=db)" --output /tmp/pantheon-db-backup.sql.gz
```
  This avoids fetching and parsing the full backup list JSON, which is large and wastes context tokens.
- Specific date requested: Only in this case, list backups to find the matching one:
```
terminus backup:list ${TERMINUS_SITE}.{environment} --element=db --format=json
```
  Filter to entries whose filename contains the target date string (e.g., 2026-02-10), then download:
```
curl -L "$(terminus backup:get ${TERMINUS_SITE}.{environment} --element=db --file={backup_filename})" --output /tmp/pantheon-db-backup.sql.gz
```

Import Database:

IMPORTANT: The db-rebuild.sh script prepends $(pwd)/ to positional file arguments (see line 80 of the script). You MUST either:
- cd /tmp before invoking the script and pass the relative filename, OR
- Pass the file using the -- separator to avoid the path prepending

Run the import:

cd /tmp && /usr/local/bin/db-rebuild.sh pantheon-db-backup.sql.gz

This script will:
- Drop and recreate the local database
- Detect compression format and decompress
- Import the SQL into the local database
- Run drush deploy to apply config and migrations
Stream the output to the user so they can see progress
This command takes a long time to run — use a generous timeout (at least 10 minutes)

Load Db | Skills Pool

Load Db

Load Db

Load Database

Instructions

Usage

Interpreting Natural Language

Steps

Error Handling

Database Migrations

Database Migrations

Postgres Patterns

Frontend Query & Mutation

Db Migrations

Drizzle