Sync content and data between local and cloud environments.

Help: { "command": "cloud sync" } via systemprompt_help Requires: Active session -> See Session Playbook

Check Cloud Status

{ "command": "cloud status" }

Check authentication:

{ "command": "cloud auth whoami" }

Sync Files to Cloud

Dry Run First

{ "command": "cloud sync push --dry-run --verbose" }

Execute Sync

{ "command": "cloud sync push --verbose" }

This syncs files from storage/, not database content.

Sync Down from Cloud

{ "command": "cloud sync pull" }
{ "command": "cloud sync pull --dry-run" }

Sync Up to Cloud

{ "command": "cloud sync push" }
{ "command": "cloud sync push --dry-run" }

Local Content Sync

Content can sync in BOTH directions using different commands.

How Content Sync Works

Disk (services/content/)
    │                    ▲
    │ publish_pipeline   │ content_sync (to-disk)
    ▼                    │
Database (markdown_content)
    │
    ▼ publish_pipeline prerenders
Static HTML (web/dist/)

Sync Disk → Database (Publish)

Run the publish pipeline to ingest content from disk:

{ "command": "infra jobs run publish_pipeline" }

This:

  1. Reads markdown files from services/content/
  2. Upserts to markdown_content table in database
  3. Prerenders to static HTML

Sync Database → Disk (Export)

Export content from database to disk files (e.g., AI-generated blog posts):

systemprompt infra jobs run content_sync -p direction=to-disk

Or for a specific source:

systemprompt infra jobs run content_sync -p direction=to-disk -p source=blog

This:

  1. Reads content from database
  2. Creates/updates markdown files in services/content/
  3. Preserves frontmatter format

Use this when: AI agents create content in the database and you want to save it to disk for version control.

Edit Content

To edit content, modify the markdown files on disk, then run publish_pipeline:

# Edit the file
vim services/content/blog/my-post/index.md

# Re-ingest to database
systemprompt infra jobs run publish_pipeline

CLI edit commands only update the database temporarily. Changes will be overwritten on next publish_pipeline run.

Skills Sync

View Diff

{ "command": "cloud sync local skills --dry-run" }

Execute Sync

{ "command": "cloud sync local skills" }

Database Comparison

Count Local Content

{ "command": "infra db query \"SELECT COUNT(*) FROM markdown_content\"" }

Count Cloud Content

{ "command": "cloud db query --profile <profile-name> \"SELECT COUNT(*) FROM markdown_content\"" }

Compare Sources

{ "command": "infra db query \"SELECT source, COUNT(*) FROM markdown_content GROUP BY source\"" }
{ "command": "cloud db query --profile <profile-name> \"SELECT source, COUNT(*) FROM markdown_content GROUP BY source\"" }

Troubleshooting

UUID parsing error: Local profile has non-UUID tenant ID. Switch to a cloud profile first with admin session switch <profile-name>.

No content synced (0 items): cloud sync push only syncs files from storage/, not database content. Content lives on disk in services/content/ and is ingested via publish_pipeline.

Cloud DB connection failed: Check profile with cloud profile show <profile-name> and ensure external_db_access: true in tenant config.

CLI edit changes disappeared: CLI edit commands only update the database. The next publish_pipeline run overwrites DB with disk content. Edit the markdown files on disk instead.

-> See Cloud Playbook for cloud authentication and profiles. -> See Deploy Playbook for deployment workflows. -> See Session Playbook for switching profiles. -> See Database Playbook for cloud database operations.

Quick Reference

Task Command
Cloud status cloud status
Auth check cloud auth whoami
Push files cloud sync push
Pull files cloud sync pull
Content: Disk → DB infra jobs run publish_pipeline
Content: DB → Disk infra jobs run content_sync -p direction=to-disk
Sync skills cloud sync local skills
Cloud DB query cloud db query --profile <name> "<SQL>"
Local DB query infra db query "<SQL>"