my-rea-toolkit: Evidence-backed Static Analysis for DeepSeek Harness
akash-digitavision/my-rea-toolkit
Gives DeepSeek Harness the REA reverse-engineering toolset as native mcp__rea__* tools for evidence-backed local static analysis of JS/Electron apps, extensions, ASAR, .NET and native binaries.
my-rea-toolkit is a DSH plugin that packages reverse-engineering capability into DeepSeek Harness, letting the agent call the REA toolset as native tools. It targets the situation of facing unfamiliar software with no clear starting point: JavaScript/Electron apps, browser extensions, ASAR archives, .NET assemblies, native binaries, SQLite databases, WASM and EVM bytecode can all be analyzed statically and locally. Every conclusion carries evidence and stated limits, target files never leave the machine, and a bootstrap discovers Node.js, verifies the bundled runtime and installs locked dependencies.
How to Install
dsh plugin --profile web add github:akash-digitavision/my-rea-toolkit- Category
- Tools & Capabilities
- Platform
- DSH Plugin
- Author
- akash-digitavision
- Distribution
- Plugin
my-rea-toolkit Key Features
my-rea-toolkit Repo Summary
What Does It Do?
REA Toolkit is a DSH plugin for DeepSeek Harness that brings the REA (Reverse Engineer Anything) toolset into the harness as native mcp__rea__* tools. It solves the problem of facing an unfamiliar piece of software with no clear starting point: JavaScript/Electron applications, browser extensions, ASAR archives, .NET assemblies, native binaries, SQLite databases, WASM and EVM bytecode can all be analyzed statically and locally. Every conclusion comes with supporting evidence and stated limitations, and target files are never uploaded anywhere.
Core Features
- Multi-format static analysis: covers JavaScript/Electron apps, browser extensions, ASAR archives, .NET assemblies, native binaries, SQLite databases, WASM and EVM bytecode.
- Evidence-backed conclusions: each finding ships with the evidence behind it and the limits of that evidence, so results can be verified rather than trusted blindly.
- Local-only by design: REA reads only the files or directories you point it at, does not execute the analyzed application, and does not send targets off the machine.
- Runtime auto-discovery: the bootstrap finds a suitable Node.js, verifies the bundled runtime, installs the locked production dependencies, and persists the runtime paths at user scope.
- Optional deep engines: Hopper, Ghidra and IDA are bring-your-own extras; pure JavaScript and extension analysis needs none of them.
How to Use This Plugin?
After enabling it in DSH, run the bootstrap once as described in the README and restart DeepSeek Harness so the REA runtime is verified and its dependencies are installed. Then simply ask the agent in natural language, for example to analyze an extracted extension directory and list its modules, IPC channels and gateway endpoints, or to inspect an Electron app's ASAR and explain how the renderer talks to the main process. Without the bootstrap the bundle still installs safely, but no REA tools appear until npm/npx is available.
How to Troubleshoot This Plugin?
If no REA tools show up after installation, the bootstrap has most likely not been run yet; complete it once per the official instructions and restart. If the bootstrap cannot find a suitable Node.js, check that the local version meets 22.19+/24.11+/26+, or let it use the runtime shipped with DSH. Native analysis lanes are limited on macOS Apple silicon per REA's documentation; fall back to the JavaScript analysis lanes when they are unavailable.
This page is an independent rewrite of the plugin's official README — for authoritative documentation and the latest changes, refer to the source: akash-digitavision/my-rea-toolkit. The plugin is third-party code that runs on your machine once installed; inclusion does not imply endorsement — please review the source before installing.