Checks
Most compliance tooling runs after development is done. By then, the violation is already in the codebase, the developer has moved on, and fixing it means context-switching back to something finished days ago.
Neural Inverse Checks runs while you write. The violation appears in the editor the moment it is introduced - the same way a type error appears before the code compiles.
How It Works
Two analysis layers run simultaneously on every save:
Static analysis evaluates rules against the AST and language server data. No LLM, no network call - sub-100ms, runs offline.
AI reasoning handles violations that require understanding what the code does, not just how it looks. Two phases: one to comprehend the framework rules, one to evaluate the file against them. Runs in the background, results appear within seconds.
Both layers write to the same feed. One view in the editor gutter. One panel showing every violation, every domain, every file.
Compliance Frameworks
Built-in frameworks:
- MISRA C:2012 and 2023 - embedded C safety
- MISRA C++ - embedded C++ safety
- CERT C - C secure coding
- IEC 62304 - medical device software
- ISO 26262 - automotive functional safety
- IEC 61508 - industrial safety-critical
- DO-178C - aerospace software
- SOC 2 Trust Services Criteria
- AUTOSAR
- Neural Inverse Enterprise Security Standard
Custom frameworks are defined in JSON and evaluated by the same pipeline. If your organisation has an internal security standard or a framework specific to your industry, it can be loaded alongside the built-in ones.
The Audit Trail
Every violation detected is written to the audit trail - not just flagged in the editor and forgotten. The record includes what rule was violated, which file and line, which developer wrote the code, which frameworks were active at the time, and what the AI reasoning concluded.
If a violation is suppressed, that suppression is logged too - with the justification and the identity of whoever approved it.
The audit trail is immutable. A compliance officer or regulator can trace the compliance history of any file back to the first commit.
External Tool Integration
Checks integrates with static analysis tools your team already uses. ESLint, SonarQube, CodeQL, Semgrep, Polyspace, MATLAB Static Analysis - configure them once, and their findings appear in the same feed alongside Checks results. One view, one audit trail, no tool switching.
Blocking Rules
Mark any rule as blocking and the code cannot be committed until the violation is resolved or explicitly overridden. Override requires a compliance officer role or the appropriate IAM permission. The override is logged.
This is how regulated teams enforce that nothing ships with a known violation in a safety-critical rule - without relying on process or manual review.
Cross-File Impact
When a violation is detected, the Checks panel shows which downstream files are affected. If a utility function has an unsafe data handling pattern, every file that imports it - directly or transitively - is shown in the impact chain. Compliance officers see the full blast radius from one screen.
Checks Agent
Ask compliance questions in plain language. The Checks Agent is a terminal-style interface that understands your codebase's violation state:
- What are the blocking violations across this project?
- Which files have the most compliance issues?
- What does this violation mean and how do I fix it?
- Run a full workspace scan and summarise the results.
The agent has structured access to the violation feed, the framework rules, and the impact graph - the same data the compliance dashboard shows, queryable in natural language.
Available in Enterprise
Checks is a Neural Inverse Enterprise feature. The open-source IDE includes the editor and AI coding tools. Enterprise adds the compliance engine, the audit trail, and the policy controls.