Continuous Governance
at Machine Speed
Architecture goes from invisible to measurable. Violations find you, not the other way around. Powered by industrial control theory and the pluggable Modulus substrate.
The Three Beats
How IArchitecture shifts codebase governance from manual checklists to automated control theory.
1. Measurement
"Architecture goes from invisible to measurable." Traditional metrics track code decay. IArchitecture vectorizes your dependency graph via a tuned DBSCAN to create anomaly clusters. These clusters are analyzed to catch coupling anomalies as an example. Many aspects of the clusters and supporting deterministic measurements are compiled into a score that cannot be gamed. [Patent Pending]
2. Enforcement
"Violations find you, not the other way around." Real-time, cross-file diagnostics highlight violations while editing. During a governance cycle, legitimate structural anomalies generate remediation proposals sent directly to the architect in your ticketing system with a complete evidence package. AI proposes, deterministic primitives check, and your architect decides. [Patent Pending]
3. Optionality
"IArchitecture is the first product built on Modulus. [Patent Pending]" Built on top of the Modulus orchestration substrate, it separates application logic from integration tasks. Swap your ticketing systems, messaging platforms, or data layers with a single configuration line with zero engine recompilation. Support migrations with parallel conduits running side-by-side.
The Closed-Loop Governance & PID Control
Codebase maintenance shouldn't rely on developers remembering rules. IArchitecture implements governance as a continuous feedback loop driven by a textbook Proportional-Integral-Derivative (PID) controller.
- The Governance Loop: The unsupervised embedding sensor scans type topology, aggregates anomalies, generates CAs, logs architect reviews, and commits git-native ledger state transitions.
- The PID Control Loop: Measures the rate of architectural change (drift). It outputs a single scalar pressure signal with active EMA input filtering, anti-windup protection, and integral decay on oscillation to prevent developer alert fatigue.
- Dynamic Actuation: The control loop modulates enforcement pressure (such as severity level and proposed issue release rates) based on the team's absorption capacity, closing the loop.
PID Governor Telemetry
Active FeedbackSystem Feedback Topography
Technology & Diligence FAQ
Answers to key technical architectural questions prepared for strategic due diligence.
Q: Isn't this just static analysis with extra steps?
No. Static analyzers are passive diagnostic tools that count syntax issues and require constant rule tuning. IArchitecture is a closed-loop control system. It detects anomalies automatically using vector topology, proposes enforcement rules dynamically, tracks rule efficacy, and actively throttles rules using PID control to manage developer capacity.
Q: How does the PID loop prevent developer fatigue?
Automated rule engines often cause "rule fighting" oscillations. IArchitecture tracks architectural drift cycle-over-cycle. The PID loop processes this signal through an EMA input filter and dampens output gain or decays the integration history when rule oscillation is detected, pacing release rates to match absorption capabilities.
Q: How does the FFI link satisfy LGPL 2.1 licensing constraints?
To keep the core platform 100% proprietary, no LGPL code (such as the Semgrep core) enters the repository or statically links to IArchitecture assemblies. The platform communicates with native binaries exclusively at runtime via P/Invoke calls. Native DLLs reside as replaceable loose files in the `ffi/` folder, meeting all copyleft substitution requirements.
Q: How is high-fidelity cross-file analysis achieved?
The language server maintains an in-memory type dependency graph updated incrementally on didChange keystrokes using Tree-sitter and Roslyn AST wrappers. Rather than re-scanning the workspace, diagnostics are re-evaluated in parallel against the graph and published via an indexed participation lookup in under 2 seconds.
Interactive Pipeline Visualizer
Click on any node in the Modulus graph below to illuminate its dependency neighborhood and isolate its inputs and outputs.