Getting Started

This guide walks you through starting a modernisation session from scratch. You'll be running Discovery within minutes.

Prerequisites

• Neural Inverse IDE installed and running
• Source project folder (your legacy codebase)
• Target project folder (empty or existing modern project)
• An LLM provider configured (cloud or local)

Step 1 — Open the Modernisation Panel

Press Cmd+Alt+M (Mac) or Ctrl+Alt+M (Windows/Linux), or run the command:

Neural Inverse: Open Modernisation

The panel opens as an auxiliary window alongside your editor.

Step 2 — Create a New Session

Click New Modernisation Project on the IDLE screen.

The wizard opens with two panes:

• Left — project folder pickers (sources and targets)
• Right — migration pattern selector

Step 3 — Select a Migration Pattern

Choose a preset pattern from the right pane, or type a custom pattern name. The pattern determines:

• The translation language pair (e.g. COBOL → TypeScript)
• The compliance framework context
• The topology defaults (1:1, 1:N, etc.)

See Migration Patterns for the full list of 60+ presets.

Step 4 — Pick Your Projects

Based on the pattern's topology, add source and target folders:

• Source — your legacy codebase (COBOL mainframe, bare-metal C, PLC project, etc.)
• Target — the modern project (can be empty; the IDE writes translated files here)

For 1:N or N:M topologies, use the + button to add multiple sources or targets.

Click Initialise Session.

The IDE writes Modernisation.inverse files to all project roots and transitions to the WIZARD → ACTIVE view.

Step 5 — Run Discovery

Stage 1 (Discovery) starts automatically after session initialisation.

The IDE scans your source project and produces:

• Language and file inventory
• Dependency graph between modules, copybooks, and packages
• Tech debt analysis (33 categories)
• Compliance pattern detection (10+ frameworks)
• Migration unit list with effort estimates and risk levels

Discovery runs progressively through 11 phases: walking → metadata → fingerprinting → grc-scan → graph → call-graph → api-surface → schema → tech-debt → pairing → complete.

Step 6 — Review the Plan

Stage 2 (Planning) generates a CPM-scheduled migration roadmap automatically after Discovery completes.

Review:

• Phase breakdown (10 phases from foundation to cutover)
• Blocker list — units that need manual intervention before translation
• Effort estimates per unit
• Critical path

Click Approve Plan → to unlock Stage 3. The plan cannot be modified after approval.

Step 7 — Migrate Unit by Unit

Stage 3 (Migration) translates each unit according to its assigned language pair profile.

For each unit:

1. Click the unit in the migration list
2. Review the AI-translated output in the two-window editor (source left, target right)
3. Edit if needed, or re-translate with custom instructions
4. Approve or skip

Units with fingerprint match ≥90% can be auto-approved. Units below 70% require compliance officer approval.

Step 8 — Validate

Stage 4 (Validation) runs equivalence checks and compliance fingerprint comparison.

Review divergences, apply equivalence test results, and obtain required sign-offs.

Step 9 — Cut Over

Stage 5 (Cutover) requires all 4 blocking readiness checks to pass before writing files to disk. Once approved, the IDE commits all translated units to the target project and exports an audit bundle.

Resuming a Session

Open any project that contains a Modernisation.inverse file. The IDE detects it automatically and restores the session at the stage you left off.

Or run:

Neural Inverse: Open Modernisation  (Cmd+Alt+M)

Keyboard Shortcuts

Commands