DevOps & Cloud · Change impact analysis
BlastRadar
Know what a change will touch.
Before a release, see every service a change can affect and the line of code that links it. After the release, confirm those services are still healthy.
$ change-impact assess --base v1
change-impact: CRITICAL (100/100) for v1..HEAD, 5 files, 1 commits
E2 +30 CH001 Destructive database migration: ALTER TABLE orders DROP COLUMN coupon_code
E4 +18 CH005 Memory limit lowered: k8s/payments.yaml 512Mi to 256Mi
reports high direct uses column coupon_code (services/reports/nightly.py:2)
web high 1 hop calls orders, whose API changed
$ change-impact verify --report impact.json
REGRESSED orders error_rate 0.002 before, 0.031 after (+1450%)
The problem
"It was a small change." A dropped column breaks a nightly report in another folder, a lower memory limit slows the checkout, a Cloudflare setting takes down the API. Nobody connected the change to what it touched.
What it does
Reads a git range, the last successful production deploy (GitHub or GitLab), a Terraform plan before apply, or changes made in a console: CDN, DNS, IAM, database settings
Maps your services from docker-compose and Kubernetes manifests, plus a simple services.json for owners, tiers, roles and domains
Finds the services that use a dropped column, a removed environment variable, a changed IAM role, domain or database, with file and line
Follows callers one and two hops away, so you see who sees the errors
21 change rules with the known way each breaks: destructive migrations, narrowed IAM, lower memory limits, fewer replicas, probes, ports, API contracts, shared libraries, major upgrades
Risk score from LOW to CRITICAL with a checklist to run before deploying
After the deploy: health checks and before/after metrics for exactly the affected services, from Prometheus, Datadog or New Relic, or numbers from any tool
Learns from your own releases: how often each kind of change was followed by a regression
Comment on GitHub pull requests and GitLab merge requests with the score, evidence, a dependency diagram and the checklist; CI gate on the riskiest changes
Deployment timeline with the risk of each production deploy
Optional Claude review for the on-call engineer, citing the evidence
Read-only, runs in your CI or on a laptop; no agent in production
How it works
- Connect the repository and, optionally, GitHub or GitLab deployments
- Add owners, tiers and health checks for the important services
- Get a report on every pull request, and a check after every deploy
Works with
Runs in your CI
# On every pull request
uses: Deepanshu18Saini/change-impact@main
# After the deploy
change-impact verify --report impact.json
Your data
- Runs in your CI or on a laptop; your code stays where it is.
- Read-only access to deployments and metrics; it never rolls back or restarts anything.
- The optional Claude review sees the assessment only, never whole files.
Who it's for
Questions
How does it know what depends on what?
From docker-compose and Kubernetes files, environment variables that point at other services, imports, SQL that names the same tables, and a short services.json for what code can't show.
What happens after the deploy?
verify checks health URLs and compares each affected service's metrics before and after, and exits non-zero on a regression so a pipeline can stop.
Related
Often used with BlastRadar
SLOStudio
SLOs and error budgetsReliability you can measure. Define each SLO in a few lines and get correct burn-rate alerts, an error budget everyone understands, and a release gate, on the monitoring you already run.
$1,500–3,000 (₹40,000–80,000 in India)View product → DevOps & CloudTestScope
Test impact selectionRun the tests that matter. Every pull request runs the tests its change can affect, each with a stated reason, and the full suite whenever a change is too broad to trust.
$1,000–2,500 (₹25,000–60,000 in India)View product → DevOps & CloudQuietPager
Alert noise reductionQuieter nights for on-call. Find the alerts that wake people for nothing, see what on-call really costs, and get the exact configuration change that fixes each one.
$800–2,000 (₹20,000–50,000 in India)View product →See BlastRadar on your own data
A 30-minute call. We run the product on your export, repository or document and walk you through the results.