Loading...
Free GitHub Action. Comments on every PR with stale-flag findings. 12 languages, every flag SDK. No procurement, no integration call, no sales rep.
Every team with 50+ flags has a graveyard. Hardcoded trues nobody dares delete. Dead else-branches that were "temporary" 18 months ago. Experiment flags from products you sunset.
You don't notice it until you're three sprints into a refactor and someone says "wait, is legacyCheckout even on?" Then a junior dev removes it. Then production breaks at 3am.
FlagShark watches your repo continuously. Every PR gets a comment. Every stale flag gets a removal PR you can review in 30 seconds. No dashboards to forget. No quarterly cleanup sprint that never happens.
Median end-to-end on repos under 100k LOC: 4.2s · 99th percentile: 11.3s · longest ever: 38s (Kubernetes monorepo)
No dashboards to install in your team's heads. No browser extension. The PR comment is the entire interface: every stale flag, every staleness reason, every "open the removal PR" link, inline on the PR your team is already reviewing.
What this means in practice: the cleanup conversation happens in the same place as the feature conversation. No context switch.
The dashboard exists for the eng manager who wants the full view across repos. Active flags. Stale flags. Cleanup PRs merged this month.
The honest take: the Action does 90% of the value at 0% of the cost. The dashboard is for when you've installed across 10+ repos.
Every Monday, FlagShark posts a single message to your team's Slack: this week's cleanup PRs merged, new stale flags detected, longest-lived flag still in production. Two-line read.
What this won't be: a stream of notifications. We compress an entire week into one message.
Every PR opened, every flag we detect, every removal that ships. Anonymized. Live.
flagshark.yml to .github/workflows/No tokens. No prod credentials. We just read the diff.Install the free Action. Get your first cleanup PR in the next hour.