Engineering decisions backed by
data, not intuition.
Vokt builds precise call graphs from compiled Go and Java code, extracts behavioral facts from intermediate representations, and computes confidence-weighted metrics — hub scores, blast radius, risk profiles, and behavioral diffs. Every metric shows its confidence and component breakdown.
Metrics derived from the call graph, not opinions
Every metric is computed from call graphs, behavioral extraction, and git history. Every metric shows its confidence profile and component breakdown. No surveys. No estimates.
Call Graph
- Hub Scorecentrality
- Fan-In / Fan-Outcallers & callees
- SCC Detectionmutual recursion
- Edge Confidenceper-edge scoring
Behavioral
- Guardspreconditions
- Mutationsstate changes
- Error Flowsfailure paths
- Side-Effect Depthpropagation
Risk
- Risk Scorecomposite 0–100
- Blast Radiusconfidence-weighted
- Diff Severityminor to critical
- Reliability RatingHIGH / MED / LOW
Git Signals
- Churn Ratechange frequency
- Bus Factorknowledge risk
- Co-Change Anomalyhidden coupling
- Commit Attributionper-function
From visibility to productivity
One platform that serves every level of the engineering organization — from understanding what you have, to governing how it changes, to making every developer faster and safer.
Architectural Visibility
Your payment module's blast radius increased 40% this quarter. OrderService went from 8 callers to 23 — it's now a hub. Hidden coupling detected: UserAuth and BillingEngine always change together but have no direct call relationship. Vokt surfaces insights like these from the call graph.
Call Graph Exploration
Explore callers, callees, hub functions, and strongly connected components across your codebase. Trace call chains through the graph. See the architectural spine of your system.
Behavioral Specs
Generated behavioral specs per function — guards, mutations, error flows, call patterns — extracted deterministically from compiled code. The foundation for every metric.
Confidence Profiles
Every edge in the call graph carries a confidence score. Every blast radius shows its reliability profile. Run vokt verify-graph on any codebase to understand analysis quality before you rely on metrics.
Governance & Risk
A function changes every sprint, has a weighted blast radius of 31, and only one person has ever touched it. Risk score: 94/100 (MEDIUM reliability — 21% of blast zone edges are DI synthetic). Quantified risk with transparent confidence.
Transparent Risk Scores
Composite risk scores (0-100) combining blast radius, hub centrality, churn rate, bus factor, co-change anomalies, and propagation depth. Every score shows its component breakdown and reliability rating.
Behavioral Diff in CI
Run vokt diff in your pipeline. Fail on drift severity with --fail-on. Per-field diffs show exactly which guards, mutations, and error paths changed — not just line-level text changes.
Work Allocation Intelligence
Blast radius 47, touches 3 hub functions, crosses auth boundary — senior work. Blast radius 3, no hubs, single module — safe to parallelize. Complexity tiers driven by data.
Developer Productivity
Production is broken in payments. Vokt shows what changed in the last 3 commits that touches the payment module, what downstream effects those changes have, and what error paths exist. Or: 42 functions have mutations but no test file — here are the top 10 by blast radius.
Context & Incident Triage
Full behavioral context for any function — callers, callees, guards, mutations, error flows. Trace call chains across files. Combine change history with effects propagation for instant incident triage.
Behavioral Drift Detection
Per-field diffs showing exactly which guards, mutations, and error paths changed. Guard removals, contract breaks, silent regressions — caught before they ship. Severity scoring from MINOR to CRITICAL.
Refactoring Safety
Rank functions by cost-to-change (blast radius * callers * mutations). See the confidence-weighted impact of any change before touching code. The web viewer lets you explore the graph interactively.
Why existing tools aren't enough
Line-level diffs miss behavioral changes
GitHub, GitLab, and Bitbucket show you what lines changed. They can't tell you that a guard was removed from a payment function, or that 23 downstream callers are now affected with weighted blast radius of 31.
Source-level parsing produces unreliable call graphs
Tools that parse source code without type information cannot resolve interface dispatch or dependency injection. The result: inflated blast radii, spurious hub detections, and metrics you can't trust. Vokt works at the compiled level where type information is complete.
No tool combines graph + behavioral + git signals with confidence
Vokt is the only tool that builds confidence-weighted call graphs from compiled code, extracts behavioral facts from intermediate representations, combines them with git history signals, and shows you the reliability of every metric.
Stop shipping behavioral regressions
See what your code actually does. Catch what tests miss. Know the blast radius before you merge.