The Problem: If you're an engineering manager or dev lead, you've probably asked these questions:
"What shipped while I was sleeping?"
"How risky was that deploy?"
"Why does this pipeline keep failing?"
"How much are we wasting on GitHub Actions?"
I kept hearing these from eng managers. The answers always required digging through GitHub, cross-referencing Slack, and manually reconstructing timelines.
So I built DeployGuard.
What is DeployGuard?
It's a CI/CD intelligence platform with five capabilities:
Deploy Guard — Every push to production gets a risk score from 0–100 based on lines changed, files touched, test results, time of day, and author failure history.
Cost Tracker — Per-repo, per-workflow CI/CD spend visibility. Automatically detects waste like flaky test reruns.
Incident Replay — When a deploy fails, an incident is auto-created with the commit, PR, workflow, and error context.
Intelligence Engine — Automatically detects recurring failure patterns, flaky pipelines, and calculates CI waste in dollars. No rules to configure.
PR Bot — Posts a DeployGuard analysis comment directly on every PR with risk score, duration, and cost. Engineers see it without leaving GitHub.
How the Risk Scoring Works
The scoring engine evaluates 5 factors, each contributing 0–20 points:
Lines changed
Files touched (config files weighted 2x)
Test results
Time of day (Friday 5pm = high risk)
Author's 30-day failure rate
A deploy that changes 500 lines across 30 files on a Friday afternoon with no tests? That's scoring 85+ (danger zone).
A small Monday morning fix with all tests passing? Score: 10 (safe).
The Intelligence Engine
This is what makes DeployGuard different from a dashboard. It actively finds problems:
Recurring Failures — Groups failures by workflow and counts occurrences. "Your deploy-production workflow failed 5 times in 7 days" with the common commit messages that triggered failures.
Flaky Pipeline Detection — Finds workflows where the status alternates between success and failure. Calculates a "flakiness score" based on how often the status flips.
CI Waste Calculation — Sums up the dollar cost of failed builds, flaky reruns, and repeated failures. Shows you exactly how much money your CI pipeline is wasting.
The PR Bot
Every PR gets an automatic comment:
MetricValueRisk Score🟢 20/100 (safe)WorkflowCIDuration33sCI Cost$0.0080
This is the highest-impact feature. Engineers see the analysis without opening another tool.
The Tech Stack
Next.js 16 + TypeScript
Supabase (PostgreSQL + Row Level Security)
GitHub App with webhooks
Stripe (billing)
Vercel (hosting)
Slack Incoming Webhooks
Total monthly cost: $0 (all free tiers).
Try It
DeployGuard is free for 1 repo. No credit card.
Go to deployguard.dev
Connect your GitHub
See your first risk score on the next push
I'm a solo developer building this under ClearFix.co. Would love feedback from anyone managing engineering teams.
Top comments (0)