From code change to risk score
in under 2 minutes.
Buildpathio integrates at the CI layer and builds your live service graph from actual traffic, config, and code — not documentation.
Four steps from install to insight
Install the GitHub App and drop a buildpath.yaml in your repo root. No manual service topology configuration. Buildpathio discovers structure from the code itself.
On every push, the scanner traverses import graphs, OpenAPI contract files, and optional Istio/Linkerd telemetry. Each data source catches a different class of dependency the others can miss.
Blast radius (40%), change volume in dependency-critical files (25%), test coverage delta (20%), and dependency depth (15%) combine into a single 0–100 risk score per affected downstream service.
The buildpath/risk-score check appears inline in the PR — before any reviewer opens the diff. Also available in Slack, PagerDuty, and CLI output.
How we build the live dependency map
Buildpathio combines three complementary data sources to build a map that reflects reality, not documentation:
- Import scanner — static analysis of service source code identifies direct package and API dependencies at the code level.
-
OpenAPI spec reader — contract scanning across all
openapi.yamlfiles detects cross-service API surface dependencies. - Service mesh trace aggregator — runtime traffic patterns from Istio or Linkerd telemetry validate and augment the static graph with actual call frequency.
How risk scores are calculated
Every risk score is a weighted composite of four signals derived from your graph and the current change delta.
| Factor | Weight | Description |
|---|---|---|
| Blast radius | 40% | Number of downstream services transitively affected |
| Change volume | 25% | Lines changed in dependency-critical files |
| Test coverage delta | 20% | Coverage decrease for affected code paths |
| Dependency depth | 15% | Graph depth from changed node to leaf services |
See your actual dependency graph — not what the docs say it is.
Free tier. GitHub App install to first risk score in under 4 minutes.