WALGuard

Who it's for

Solo/small-team backend devs running SQLite as their primary production DB (Litestream, Turso, LiteFS, fly.io volumes users) who today only run manual 'PRAGMA integrity_check' occasionally or trust backups blindly

The problem

time + business risk: silent WAL corruption (like the 16-year-old bug Tailscale just traced) can propagate into every backup for weeks before anyone notices, costing days of incident response and possible data loss

How to build it

single static binary/sidecar agent that opens the SQLite file read-only, runs scheduled integrity_check + WAL frame/checksum diffing, pushes alerts via webhook/Slack/email, plus a small hosted dashboard showing history per database

How it makes money

$15-40/month per monitored database instance, paid by small SaaS teams and self-hosters who already pay for backup tooling and treat data-loss prevention as cheap insurance; free because DIY integrity_check exists but nobody schedules/monitors/alerts on it reliably at 3am

Why it doesn't exist yet

incumbents (Litestream, Turso, cloud DB vendors) focus on replication/backup, not on proving the live file isn't already corrupt — verifying WAL health requires deep SQLite internals knowledge most teams don't have time to build in-house, so the gap is a narrow, unsexy monitoring tool nobody big wants to own

First users

post the Tailscale WAL bug postmortem alongside a working detector as a HN/Show HN launch — devs who just read the Tailscale story will self-select to check their own instances immediately

Build size

2 people x 10 weeks: agent binary + integrity_check scheduler + WAL frame diffing + webhook/email alerts + minimal dashboard; excludes automated repair/restore and non-SQLite databases

Biggest risk

if Turso or Litestream ships built-in corruption/integrity alerting as a first-party feature, the standalone tool becomes redundant overnight

Conditions for a hit (all 3 required)

• Runs PRAGMA integrity_check on a schedule (configurable, default 15min) against a target SQLite file and logs pass/fail with timestamp
• Detects when a WAL file exceeds a configurable size threshold without a checkpoint completing, and fires a 'stuck WAL' alert within one polling cycle
• Produces a weekly emailed report listing every anomaly detected (failed checks, stuck WALs, checksum mismatches) for that period, with zero anomalies producing a clean 'all green' report

How it's judged (in 6 months)

GitHub repo reaching 500+ stars OR Product Hunt daily top 5(judgment date 2027-02-13)

AI self-confidence 42/100 — self-reported likelihood of meeting the criterion, not a business success rate