Skip to content

Security Center

From V8.5.0, rConfig checks its own security every day. About 100 read-only checks look at the application settings, the web server, TLS, PHP, the database, Redis, accounts, API tokens, stored credentials, the operating system and the security advisories that apply to your version. A check that newly fails sends an alert, and the results are on Settings > Security Center.

Running an old PHP build with a published vulnerability? An .env file reachable over HTTPS? An administrator still on admin@domain.com / admin? Security Center finds each of these and tells you the exact command to fix it.


Each check has a stable ID such as redis.auth or web.env_exposed, a severity, and a result:

Result Meaning
Passed The setting is safe.
Warning Worth fixing, but not an open door on its own.
Failed Fix this. Failures send an alert.
Skipped The check does not apply here (for example a Vector check without Vector), or could not run.
Accepted An administrator accepted the risk until a date, with a reason.

The checks are grouped. The check reference explains every ID:

  • Application, Web exposure, TLS, PHP, Database, Redis
  • Accounts and sign-in, API and MCP tokens, Data at rest
  • Runtime and privileges, Versions, Security advisories, Supply chain
  • Operating system, Integrations, Devices

The checks never change anything, and never show a secret. They report whether a value is set and safe, not the value.


Open Settings > Security Center (first under Security & Access) and click Run checks. The checks run in the background and take under a minute. You can leave the page; the results are stored.

The scheduler runs a full check every day at 04:15 (set SECURITY_CHECK_TIME in .env to change it). If the scheduler stops, runtime.security_check_stale warns after 48 hours.


The tiles at the top count failed, warning, passed, skipped and accepted checks. Click a tile to filter the list. By default the list shows what needs attention: failures first, then warnings, sorted by severity.

Click a finding to see why it matters, the fix for your server, the evidence the check collected (never a secret), the standards it maps to (NIST SP 800-53, NIS2 Article 21, CIS Benchmarks, OWASP ASVS) and any advisory it guards against.

rConfig Security Center page with posture tiles, an active advisory and an expanded failing check Security Center page with posture tiles, an active advisory and an expanded failing check showing its fix command.

The advisories card lists rConfig security advisories that apply to this server, and why: for example PHP 8.4.27 is in the affected range (>=8.4.0 <8.4.28). It links to the advisory and to all rConfig security advisories. See Security advisory feed for how this works offline.


Some findings can be fixed from the page, as the web user, without root:

Check Fix
data.config_file_modes Sets saved configs to 0440 and their folders to 0750 (rconfig:set-config-permissions).
app.env_file_perms Removes other-user access from .env, when the web user owns it.
app.env_backups Makes .env backups 0640.
web.public_dir_clean Moves backups, dumps and keys out of public/ into storage/app/quarantine. Nothing is deleted.
tokens.expired_present Deletes expired API tokens.
tokens.unused Revokes the long-unused tokens you tick.

Click the fix on the finding. rConfig shows exactly what will change, runs the fix in the background, records it in the activity log and runs the checks again. Fixing needs the security.update permission.

The same fixes are available on the command line:

Terminal window
php artisan rconfig:security-fix --list
php artisan rconfig:security-fix data.config_file_modes --dry-run
php artisan rconfig:security-fix data.config_file_modes

Everything else shows the command to run. Fixes that need root point to the server check script.


Sometimes a finding is expected. A local break-glass administrator, for example, will always show in accounts.mfa. Click Accept risk, give a reason (at least 10 characters) and a review date at most a year away.

An accepted finding stays in the report and in exports, marked Accepted. It no longer counts as a failure or warning, no longer changes the command’s exit code and no longer alerts. When the date passes it counts again. Click Stop accepting to end it early. Both actions are recorded in the activity log.


A check that newly fails sends a Security Check Failed notification, and a recovery sends Security Check Recovered. Both use your notification channels: in-app, email, Slack, Teams and webhooks. Set who gets them in Settings > Notifications. A failure alerts once, not every day.

To alert on new warnings too, set SECURITY_CHECK_NOTIFY_ON_WARN=true in .env. The check integrations.alert_channels warns if alerts would only reach the in-app bell.


  • Export > JSON gives the same report as rconfig:security-check --json (schema version 1).
  • Export > CSV gives one row per check, for a spreadsheet or an audit pack.
  • GET /api/v2/security/review returns the latest report to a REST API token with the View the Security Center report (security.read) ability, for SIEM and compliance tools.
  • The get_security_posture MCP tool gives AI assistants the same report. See MCP server.

Ask the Security Assistant to explain a finding, plan the fixes or write an audit summary.