Every tool your architecture needs
Five review modes, 20 compliance frameworks, and deep integration with your existing workflow.
Find responsibility violations before they become tech debt
The full pipeline: scan your repository structure, analyze responsibilities with AI, design abstractions, and generate test contracts. You get a migration plan with phased adoption.
- Fan-out/fan-in evidence for every violation
- AI-proposed interface abstractions with TypeScript signatures
- Generated test contracts with edge cases
- Phased migration plan with dependency ordering
Lightning-fast PR quality gates
20 structural checks run in under 5 seconds. Smart early exit skips LLM when PRs are clean — zero cost for well-written code.
- Secrets, breaking changes, test coverage, complexity, coupling
- Early exit: skip LLM if score ≥90, all checks pass, <100 lines
- Suggested reviewers with expertise areas
- Health delta: see what degraded vs improved
20 compliance frameworks, zero LLM cost
Pure structural analysis with configurable risk factor weights and severity thresholds. Get compliance coverage across privacy, security, AI governance, and industry regulations.
Risk Audit configuration in the web UI
Six checks. One score. Zero cost.
All checks run locally using CKB structural analysis — no LLM calls, no API keys needed.
Dead Code
SCIP-based unreferenced symbol detection
Test Gaps
Untested functions ranked by complexity
Complexity
Cyclomatic + cognitive per function
Coupling
Co-change file pairs with strength ratings
Secrets
26 credential patterns scanned
Cycles
Circular module dependency detection
Your complete refactoring roadmap
From hotspot identification to phased migration plans — with 6 AI reviewer personas providing specialized perspectives.
Analysis
- 01 Hotspot identification (churn x complexity)
- 02 Temporal coupling from git history
- 03 Bus factor — knowledge concentration risk
- 04 DB coupling analysis
- 05 Architecture conformance checks
AI Personas
Each persona reviews with specialized focus areas and domain-specific prompts.
Code Smells
86.7% recall with T-shirt sizing (S/M/L/XL)
"What If" Simulator
See the blast radius before you refactor. Simulate extracting a module and understand every dependency that breaks.
Review history with comparison across runs
Quality gates, architecture conformance rules, and custom rules