HOME / FEATURES
Features
Every capability FaultPilot ships today — organized the way the engine is organized. No roadmap theater: if it is on this page, it works.
Intelligent Error Engine
FaultPilot captures the failure surfaces that actually break WordPress sites and normalizes them into a single signal stream — every event fingerprinted, deduplicated and sampled so a flood of errors becomes one clean record instead of a log firehose.
- PHP errors — fatals, warnings, notices and deprecations from the error handler and shutdown
- Database errors — query failures with statement type and first table (never full SQL payloads)
- REST failures — failed route callbacks across the REST API
- Outbound HTTP failures — API calls that fail or time out, attributed to the caller
- Cron problems — missed schedules, duplicate events, Action Scheduler backlog
- Mail failures — wp_mail failures that silently swallow password resets and orders
- Admin JavaScript errors — console errors that make wp-admin misbehave
TypeError str_replace(): arg #2 (string) …
TypeError str_replace(): arg #2 (string) …
↓ normalize · fingerprint · merge hits
fp: FP-82A19C · type fatal · component woocommerce-emailer
hits: 37 merged (2 sampled) · first 14:32:08 · ongoing
Incident management & root cause analysis
Related errors merge into incidents with a real lifecycle — open, investigating, resolved, reopened. Each incident carries its evidence: affected component, recent changes, occurrence patterns and related failures.
The Root Cause Analyzer weighs that evidence — change/component match, temporal proximity, occurrence spikes — and states a primary suspect with an explicit confidence tier: Strong correlation, Likely cause, Possible cause, or Insufficient evidence. It shows the factors behind the verdict, so you can disagree with data instead of hunches.
Prevention engine & explainable risk
Hourly sweeps watch the trends that precede real failures: rising error rates, cron backlog growth, autoloaded option bloat, memory trajectory, REST and mail failure streaks, SSL expiry, disk pressure and checkout deterioration. When a trend crosses its threshold, you get a preventive alert — before the incident, not after it.
The risk model produces six component scores — reliability, updates, database, performance, compatibility, security — and every factor behind every score is visible. You can always ask why the number moved.
Sixteen real checks — each honest about its limits
Run manually, during onboarding, or daily. Each check returns a measured result or an explicit “not observable” — never a guess dressed up as data.
PHP version
Installed version vs active support and security support windows.
Memory limit
Current limit and measured headroom under real traffic.
Disk space
Remaining capacity on the site’s storage volume.
TLS certificate
Certificate expiry window and scheme health.
REST loopback
Whether the site can actually reach its own REST API.
Autoloaded options
Size and top offenders of the always-loaded option set.
Transients
Expired and orphaned transient buildup.
Table sizes
Largest tables and growth signals.
Cron health
Overdue, duplicate and stuck scheduled events.
Action Scheduler
Pending and failed background jobs.
Error backlog
Open incident and event volume by severity.
Plugin compatibility
Version signals against the current WordPress core.
Plugin health
Error attribution per plugin across recent incidents.
Theme health
Theme errors and override sanity.
WooCommerce
Checkout reliability, overrides, gateways, webhooks, HPOS signals.
Security signals
Measured hardening signals — file editing, debug exposure.
Database, Performance & Cron Doctors
Each “doctor” is a specialist view with a strict workflow: safe analysis first, explanation second, snapshot before action, explicit confirmation, then verification after execution. The Database Doctor never deletes table data automatically — you decide, with evidence in front of you.
- Database Doctor — table sizes, autoloaded option top offenders, transient buildup, orphaned meta
- Performance Doctor — memory headroom, autoload weight, measured slow cues
- Cron Doctor — duplicate events, stuck schedules, missed runs, Action Scheduler depth
✓ measured: 812 KB across 1,204 rows
$ explain --option fp_cache_blob
→ written by cache plugin, 512 KB, updated hourly
$ snapshot --label before-cleanup ✓
$ treat --option fp_cache_blob --action reset
confirm? requires explicit confirmation
$ verify ✓ autoload now 300 KB · health re-measured
API & Webhook Doctor
Modern WordPress sites depend on dozens of outbound calls — payment gateways, CRM sync, webhooks, license checks. When any of them fail, the symptom usually shows up somewhere else entirely: a stuck order, a missed email, a queue that never drains.
The API/Webhook Doctor attributes outbound HTTP failures and webhook problems to the caller, correlates them with incidents, and shows failure rates and latency patterns per endpoint — so “checkout sometimes hangs” becomes “gateway X fails 8% of the time since Tuesday’s update”.
Change timeline & update safety
Most WordPress incidents start with a change. FaultPilot records plugin, theme and core updates, activations and config changes on a timeline — and takes an automatic recovery point immediately before every update. When RCA correlates an incident with a change, the pre-update point is already there.
- Automatic pre-update snapshots of the plugin set
- One-click rollback to a captured point
- Config change tracking — permalink changes, option flips, activation events
Smart alerts, reports & activity logs
Alerts reach you without drowning you: deduplication merges repeated signals, throttling prevents alert storms, minimum-severity filters keep the noise down, and quiet hours respect your nights. The dashboard notification center and a branded HTML email channel share the same rules.
Every incident, treatment and recovery action is written to an audit-logged activity timeline. Structured and printable reports export the full picture for clients and post-mortems.
AI-ready architecture
A stable, structured diagnostic-context endpoint exposes the site’s current picture — incidents, risk factors, diagnostics — for a future AI layer. No external AI service is required for any capability, and no data leaves the site unless you decide to send it somewhere.
Security & privacy by design
Dedicated capability, nonce-validated state changes, prepared SQL, redacted sensitive keys, hashed recovery tokens, data minimization and no unnecessary external transmission. The full threat model is public.
Explore every screen
The interactive demo shows the dashboard, incident views, doctors and recovery flows with realistic data.