Claude Code Plugin

This page covers the Claude Code plugin that powers the reverse engineering process. The plugin provides a set of skills (slash commands) and agents (autonomous subagents) that together analyse legacy applications and produce structured deliverables.

Prerequisites

You must have Claude Code installed and authenticated before using the plugin.

See the official Claude Code documentation: https://docs.anthropic.com/en/docs/claude-code

This plugin is not published on the marketplace yet, and in the interim, you can clone the plugin from its GitHub repository:

git clone https://github.com/DEFRA/claude-legacy-reveng-plugin

Running with the plugin

Launch Claude Code with the plugin directory:

claude --plugin-dir /path/to/claude-legacy-reveng-plugin

Or set up a shell alias for convenience:

alias claude-lap='claude --plugin-dir /path/to/claude-legacy-reveng-plugin'

Skills

Skills are slash commands that perform focused, single-purpose tasks.

Skill Description
image-to-html Converts a legacy UI screenshot into semantic, unstyled mockup HTML
curate-transcript Removes off-topic content from interview transcripts whilst preserving domain knowledge verbatim
validate-mermaid Validates all Mermaid diagram blocks in a markdown file and fixes broken diagrams in place
prd-to-features Decomposes a PRD into individually deliverable feature specifications with user stories, wireframes, and acceptance criteria

Agents

Agents are autonomous subagents that carry out broader analysis tasks.

Agent Description
digital-content-curator Prepares raw screenshots and transcripts into structured, analysis-ready outputs
business-analyst Extracts strategic DDD patterns from curated transcripts and HTML mockups
interaction-analyst Produces screen inventory, user workflows, and navigation maps from mockups and transcripts
application-developer Reads legacy .NET source code to extract workflows, behaviours, domain model, and business rules
database-analyst Reads SQL Server database code to extract schema, stored procedures, and database-level business rules
product-manager Orchestrates all analysts and synthesises a comprehensive PRD

Component map

The following diagram shows how skills and agents relate to one another.

graph LR
    subgraph Skills
        i2h
        ct
        vm
        p2f
    end

    subgraph Agents
        curator[digital-content-curator]
        ba[business-analyst]
        ia[interaction-analyst]
        ad[application-developer]
        da[database-analyst]
        pm[product-manager]
    end

    curator -->|invokes| i2h
    curator -->|invokes| ct

    pm -->|orchestrates| ba
    pm -->|orchestrates| ia
    pm -->|orchestrates| ad
    pm -->|orchestrates| da

    ba -->|invokes| vm
    ia -->|invokes| vm
    pm -->|invokes| vm

    p2f -->|reads| prd[output/PRD.md]
    p2f -->|writes| ft[output/features/FT-*.md]

Further information

Refer to the plugin repository for full documentation, including detailed agent and skill definitions.