Audit. Reproduce. Repair.
Follow each issue from initial signal to verified patch, backed by automated tests and ready for your team’s code review.
From repository audit to verified resolution.
Rehabilty goes beyond static warning lists: issues are reproduced in an isolated sandbox, repaired with targeted patches, and verified with automated tests.
- 01
Repository Audit & Risk Mapping
Connect your GitHub, GitLab, or Bitbucket repository. Rehabilty runs deep static and semantic analysis to surface security vulnerabilities, performance bottlenecks, and architectural defects.
INPUT → MULTI-LAYER SCAN - 02
Isolated Sandbox Reproduction
Unlike ordinary linters that drown you in false alarms, Rehabilty spins up an isolated Docker container and executes live reproducer test cases to verify if the vulnerability is genuinely exploitable.
EXECUTION → LIVE REPRODUCER PROOF - 03
Surgical Patch & Regression Verification
Rehabilty writes minimal, targeted code repairs — fixing root causes without touching unrelated code. The patch is verified in the sandbox against project regression suites with zero side effects.
VERIFICATION → 100% REGRESSION PASS - 04
Dedicated Git Branch & Zero Risk to Main
Rehabilty never touches or overwrites your main branch. All verified code changes, execution logs, and before/after diffs are pushed to a dedicated branch (rehabilty/verified-repairs) ready for developer review.
DELIVERY → REVIEWABLE GIT BRANCH
Technical architecture and operating principles.
Input & Deliverables Model
The input is a Git repository and an agreed audit scope. The output is a prioritized finding report and a dedicated Git branch containing verified, regression-tested code changes. Rehabilty is not a prompt-to-app generator; it is an automated audit and repair system designed for existing production codebases.
Security-First Review Scope
Security is the primary directive: broken access controls, unsafe input handling, exposed credentials, and vulnerable dependencies. Audits also cover maintainability, performance bottlenecks, and accessibility standards. Tooling is configured automatically based on repository stack, and each report explicitly details what was verified.
Isolated Sandbox Runtime Testing
The application runs inside an isolated, secure container where realistic user flows and exploit payloads can be tested safely. Hypotheses from static analysis are confirmed against runtime behavior. Findings are backed by concrete execution logs and reproducible test cases, without any impact on production environments.
Root-Cause Repair & Regression Verification
The repair engine analyzes error logs and relevant source code to synthesize a targeted patch. Both the failing test and existing regression suites are executed against the modified code. A repair is only accepted when all verification assertions pass with zero side effects. Unresolved issues or decisions requiring developer judgment are explicitly flagged in the report.
Developer Ownership & Branch Handoff
All accepted changes are committed to a dedicated Git branch with complete before-and-after diffs, execution logs, and test evidence. The default branch is never modified automatically; your engineering team retains total ownership of code review and merge decisions.
Role of AI Models & Deterministic Tooling
Language models provide code reasoning, vulnerability analysis, and patch synthesis. However, model output is never accepted as proof on its own; every proposal must pass deterministic static analyzers and executable integration tests. A multi-model pipeline routes complex architectural decisions to high-capability models and bounded tasks to fast, focused models.
Current status: Rehabilty is under active development. The public sample report demonstrates the end-to-end verification format; hosted repository access is rolling out progressively.