administration

name	administration
description	n8n instance administration — data tables, folder management, community packages, Docker deployment, environment variables, backup/restore, upgrades, CLI commands, and troubleshooting. Operations the MCP does NOT cover. Auto-triggers on n8n data tables, n8n folders, move workflow, community packages, n8n setup, n8n upgrade, n8n backup, and n8n instance management tasks.
user-invokable	true

n8n Administration

Operations the MCP and public API don't fully cover: data tables, folders, community packages, infrastructure, backup, CLI.

Source: n8n-io/n8n GitHub repo, master branch.

---

When to Use What

You want to...	Use	Why
Build/edit/validate workflows	MCP	Partial updates, validation, auto-fix
Search nodes, get node schema	MCP	Indexed node database with properties
Deploy templates	MCP	Auto-fix, version upgrades
Test/trigger workflows	MCP	Auto-detects trigger type
Manage executions	MCP	List, get, delete with filtering
Health check	MCP	n8n_health_check
CRUD credentials, test credentials	See credentials skill	Full credential management
CRUD data table rows	Public REST API	See section 1 below
CRUD data table columns, CSV import	Internal API	See section 1 below
Manage folders, move workflows	Internal API	Enterprise only — see section 2
Install community packages	Internal API	See section 3
Export decrypted credentials	CLI	n8n export:credentials --decrypted
Full database backup/restore	CLI	n8n export:entities / import:entities
Rollback database migration	CLI	n8n db:revert
Set encryption key, DB config	Environment variables	Not exposed via any API
Upgrade n8n version	Docker	Pull new image, redeploy

Priority: MCP → Public API → Internal API → CLI → Docker/Infra.

---

1. Data Tables

n8n's built-in SQLite-backed structured storage. Each table belongs to a project.

Source: data-table.controller.ts, data-table-aggregate.controller.ts, data-table-uploads.controller.ts.

Data Model

• Table: id (UUID), name (1-128 chars, unique per project), projectId, columns, createdAt, updatedAt
• Column: id, name (1-63 chars, regex ^[a-zA-Z][a-zA-Z0-9_]*$, unique per table), type (string | number | boolean | date), index (position)
• Row system columns (auto-managed): id (auto-increment int), createdAt, updatedAt

Filter Schema

Used by get rows, update rows, delete rows, upsert:

{
  "type": "and",
  "filters": [
    { "columnName": "status", "condition": "eq", "value": "active" },
    { "columnName": "score", "condition": "gte", "value": 80 }
  ]
}

Conditions: eq, neq, like, ilike, gt, gte, lt, lte. like/ilike auto-wraps with % if no wildcards in value.

Public API (API key auth)

Method	Path	What
GET	/api/v1/data-tables	List tables. Query: limit, cursor, filter (JSON: name, projectId), sortBy
POST	/api/v1/data-tables	Create table. Body: { name, columns: [{ name, type }] }
GET	/api/v1/data-tables/:id	Get table with columns
PATCH	/api/v1/data-tables/:id	Rename. Body: { name }
DELETE	/api/v1/data-tables/:id	Delete table
GET	/api/v1/data-tables/:id/rows	Get rows. Query: limit, cursor, filter, sortBy, search
POST	/api/v1/data-tables/:id/rows	Insert rows. Body: { data: [...], returnType?: "count"|"id"|"all" }
PATCH	/api/v1/data-tables/:id/rows/update	Update by filter. Body: { filter, data, returnData?, dryRun? }
POST	/api/v1/data-tables/:id/rows/upsert	Upsert. Body: { filter, data, returnData?, dryRun? }
DELETE	/api/v1/data-tables/:id/rows/delete	Delete by filter. Query: filter (required), returnData, dryRun

Internal API additions (session cookie)

Base: /rest/projects/:projectId/data-tables

Method	Path	What
POST	/:id/columns	Add column. Body: { name, type, index? }
DELETE	/:id/columns/:colId	Delete column
PATCH	/:id/columns/:colId/move	Reorder. Body: { targetIndex }
PATCH	/:id/columns/:colId/rename	Rename. Body: { name }
GET	/:id/download-csv	Download CSV. Query: includeSystemColumns
POST	/:id/import-csv	Import CSV. Body: { fileId }

CSV upload (multipart): POST /rest/data-tables/uploads — field file + hasHeaders ("true"/"false"). Returns { id, rowCount, columnCount, columns }. Use returned id as fileId in import-csv or table creation.

Global listing: GET /rest/data-tables-global/ — all tables across projects. Storage limits: GET /rest/data-tables-global/limits — total bytes, quota status.

Data Tables Gotchas

1. Filter mandatory for delete — prevents accidental full deletion
2. dryRun: true available on update, upsert, delete — previews changes without persisting
3. Column names must match ^[a-zA-Z][a-zA-Z0-9_]*$, max 63 chars
4. Table names unique per project — 409 on conflict
5. Column CRUD is internal API only — not in public API
6. CSV import is two-step: upload file → use fileId
7. Public API creates tables in API key owner's personal project
8. Source control protected instances block all writes (403)
9. Dates stored as UTC ISO 8601

---

2. Folder Management

Organize workflows into nested folders within projects.

Source: folder.controller.ts, workflows.controller.ts, folder.schema.ts.

Enterprise-only — all endpoints require @Licensed('feat:folders'). Internal API only — no public API endpoints for folders.

Folder Endpoints

Base: /rest/projects/:projectId/folders (session cookie auth)

Method	Path	What
POST	/	Create. Body: { name, parentFolderId? }
GET	/	List. Query: skip, take, sortBy, filter (JSON: parentFolderId, name, tags)
GET	/:id/tree	Ancestor chain from root to this folder
GET	/:id/content	Recursive counts: { totalSubFolders, totalWorkflows }
GET	/:id/credentials	Credentials used by workflows in this folder
PATCH	/:id	Update. Body: { name?, tagIds?, parentFolderId? } — setting parentFolderId moves the folder
DELETE	/:id	Delete. Query: transferToFolderId (optional)
PUT	/:id/transfer	Transfer to another project. Body: { destinationProjectId, destinationParentFolderId, shareCredentials }

Folder Name Validation

• Cannot be empty
• Cannot contain: [ ] ^ \ / : * ? " < > |
• Cannot consist of only dots or start with a dot
• Max 128 characters

Moving Folders

Set parentFolderId in PATCH body:

• Cannot set a folder as its own parent
• Cannot move into own descendants (circular reference check)
• New parent must be in same project
• Use PROJECT_ROOT to move to top level (sets parentFolder to null)

Deleting Folders — Two Modes

1. Without transferToFolderId: Flatten + archive — all workflows moved to project root, archived and deactivated, then folder deleted
2. With transferToFolderId: Transfer — all children moved to target folder (or PROJECT_ROOT), preserves workflow state, then folder deleted

Moving Workflows Between Folders

On the workflows controller, not the folder controller:

Operation	How
Create in folder	POST /rest/workflows/ with parentFolderId in body
Move to folder	PATCH /rest/workflows/:id with parentFolderId in body
Move to root	Set parentFolderId to PROJECT_ROOT or null
List in folder	GET /rest/workflows/?filter={"parentFolderId":"..."}&includeFolders=true
Transfer + place	PUT /rest/workflows/:id/transfer with destinationParentFolderId in body

---

3. Community Packages

Source: community-packages.controller.ts. Internal API only, session cookie auth.

Method	Path	What
POST	/community-packages	Install. Body: { "name": "n8n-nodes-package-name" }
GET	/community-packages	List installed
PATCH	/community-packages	Update. Body: { "name": "n8n-nodes-package-name" }
DELETE	/community-packages	Uninstall. Body: { "name": "n8n-nodes-package-name" }

Packages persist in the volume at /home/node/.n8n/nodes/. If volume is lost, reinstall.

---

4. CLI Commands

Run inside the container: docker exec -it <container> n8n <command>.

Backup & Export

Command	What
n8n export:workflow --all --output=file.json	Export all workflows
n8n export:workflow --id=ID --output=file.json	Export one workflow
n8n export:credentials --all --decrypted --output=file.json	Export all credentials in plaintext
n8n export:entities --outputDir=dir/	Full database export

Restore & Import

Command	What
n8n import:workflow --input=file.json	Import workflows
n8n import:credentials --input=file.json	Import credentials
n8n import:entities --inputDir=dir/ --truncateTables	Full restore (destructive)

Instance Management

Command	What
n8n execute --id=ID	Run workflow without webhook
n8n update:workflow --id=ID --active=true	Activate
n8n update:workflow --all --active=false	Deactivate all
n8n list:workflow	List all
n8n delete:workflow --id=ID	Delete
n8n audit	Security audit
n8n db:revert	Rollback last DB migration

User & Auth

Command	What
n8n user-management:reset	Nuclear reset to initial setup
n8n mfa:disable --email=[e-mail]	Disable MFA
n8n ldap:reset	Clear LDAP config
n8n license:info	License info
n8n license:clear	Remove license

Worker Mode

Command	What
n8n start	Main process (editor + triggers)
n8n worker --concurrency=5	Worker (queue mode)
n8n webhook	Dedicated webhook process

---

5. Docker Deployment

PostgreSQL (production)

version: "3.8"
services:
  n8n:
    image: docker.n8n.io/n8nio/n8n:latest
    restart: unless-stopped
    ports:
      - "5678:5678"
    environment:
      - N8N_HOST=${N8N_HOST}
      - N8N_PROTOCOL=https
      - WEBHOOK_URL=https://${N8N_HOST}/
      - N8N_ENCRYPTION_KEY=${N8N_ENCRYPTION_KEY}
      - DB_TYPE=postgresdb
      - DB_POSTGRESDB_HOST=postgres
      - DB_POSTGRESDB_PORT=5432
      - DB_POSTGRESDB_DATABASE=n8n
      - DB_POSTGRESDB_USER=n8n
      - DB_POSTGRESDB_PASSWORD=${DB_PASSWORD}
      - EXECUTIONS_DATA_PRUNE=true
      - EXECUTIONS_DATA_MAX_AGE=168
      - GENERIC_TIMEZONE=${TIMEZONE}
    volumes:
      - n8n_data:/home/node/.n8n
    depends_on:
      - postgres

  postgres:
    image: postgres:16
    restart: unless-stopped
    environment:
      - 
       
       
    
       


  
  

SQLite (simple)

services:
  n8n:
    image: docker.n8n.io/n8nio/n8n:latest
    restart: unless-stopped
    ports:
      - "5678:5678"
    environment:
      - N8N_HOST=${N8N_HOST}
      - WEBHOOK_URL=https://${N8N_HOST}/
      - N8N_ENCRYPTION_KEY=${N8N_ENCRYPTION_KEY}
      - DB_TYPE=sqlite
    volumes:
      - n8n_data:/home/node/.n8n

---

6. Key Environment Variables

Critical

Variable	Purpose
N8N_ENCRYPTION_KEY	Encrypts credentials. Irreplaceable.
WEBHOOK_URL	Full public URL. Must end with /.
DB_TYPE	sqlite or postgresdb
GENERIC_TIMEZONE	Default for cron triggers

Execution

Variable	Default
EXECUTIONS_DATA_PRUNE	false
EXECUTIONS_DATA_MAX_AGE	336 (hours)
N8N_CONCURRENCY_PRODUCTION_LIMIT	-1 (unlimited)
N8N_EXECUTIONS_MODE	regular (or queue for Redis workers)

Misc

Variable	Default
N8N_LOG_LEVEL	info
N8N_COMMUNITY_PACKAGES_ENABLED	true
N8N_RUNNERS_ENABLED	false

---

7. Backup & Restore

Quick Backup

docker exec n8n n8n export:workflow --all --output=/home/node/.n8n/backups/workflows.json
docker exec n8n n8n export:credentials --all --decrypted --output=/home/node/.n8n/backups/credentials.json
docker cp n8n:/home/node/.n8n/backups/ ./n8n-backup-$(date +%Y%m%d)/

Database Backup

# PostgreSQL
docker exec postgres pg_dump -U n8n n8n > n8n-db-$(date +%Y%m%d).sql

# SQLite
docker cp n8n:/home/node/.n8n/database.sqlite ./n8n-db-$(date +%Y%m%d).sqlite

Restore

docker exec n8n n8n import:workflow --input=/home/node/.n8n/backups/workflows.json
docker exec n8n n8n import:credentials --input=/home/node/.n8n/backups/credentials.json
# Full restore (DESTRUCTIVE):
docker exec n8n n8n import:entities --inputDir=/dir/ --truncateTables

---

8. Upgrades

docker pull docker.n8n.io/n8nio/n8n:latest
docker compose down && docker compose up -d

Pin version: image: docker.n8n.io/n8nio/n8n:1.70.3

Rollback: docker exec n8n n8n db:revert

---

9. Troubleshooting

Symptom	Cause	Fix
Webhooks not receiving	Wrong WEBHOOK_URL	Set to full URL with trailing /
Credentials invalid after move	Different encryption key	Use original N8N_ENCRYPTION_KEY
Schedules not triggering	Wrong timezone	Set GENERIC_TIMEZONE
Community nodes missing	Volume not persisted	Check volume mount
"Workflow could not be activated"	Missing credentials	Re-link credentials

Health: curl -s "$N8N_URL/healthz" or MCP n8n_health_check. Debug: N8N_LOG_LEVEL=debug then docker logs n8n -f --tail 100.

---

10. Gotchas

1. Encryption key is irreplaceable
2. WEBHOOK_URL must end with /
3. Test vs production webhooks — external systems use /webhook/, not /webhook-test/
4. Credential IDs are instance-specific — relink after import
5. Webhook paths must be unique across ALL workflows
6. Internal API requires session cookie — not API key
7. Folders are enterprise-only — requires feat:folders license
8. Data table column CRUD is internal API only