The graph-first analyst
Analyse required behaviour
The engine strategically decides what to reverse and plans out evidence to pass to the final answer writer, resulting in faster, cheaper and more accurate answers.
AI reverse engineering for IDA Pro, Ghidra, and Binary Ninja
Reverse engineering,
supercharged.
A conversational, autonomous parallel analysis tool for reverse engineering.
Product demo video
Video demo coming soon.
Final hosted demo embed URL is pending.
Why Decompile.re
Not another AI agent.
All other AI reverse-engineering plugins are agents: a single model tool-calling through the binary one step at a time. That approach is slow, costly, and loses coherence as the call graph and context grow. Decompile.re uses an LLM-powered recursive analysis engine instead: it analyses functions in parallel, tracks findings structurally, and builds a coherent program-level view before producing the final answer, effectively mitigating all of these risks.
Agent / MCP plugins
Decompile.re
Analysis method
Agent / MCP plugins
Inefficient LLM tool-calls every step: fetch, read, rename.Decompile.re
Structured pipeline, no tool-calls until the final report.Speed
Agent / MCP plugins
Serial round-trips, one action at a time.Decompile.re
Functions analysed in parallel across the call tree.Concurrency
Agent / MCP plugins
Single agent, one function at a time.Decompile.re
Multi-threaded across independent branches.Reasoning
Agent / MCP plugins
Whole-context reasoning clouding LLM memory and adding to growing context leading to slower, more expensive, and worse answers.Decompile.re
Structured context reasoning: only relevant information is passed to each function context resulting in faster, cheaper and more accurate reversal.Refinement
Agent / MCP plugins
Backwards refinement is rare and often leads to excessive LLM usage.Decompile.re
Bottom-up passes revisit callers once callees are understood.Virtual calls
Agent / MCP plugins
Indirect and virtual calls become dead ends.Decompile.re
C++ vtable and virtual-dispatch resolution are part of the workflow.Product
The graph-first analysis engine.
Autonomous analysis engine which structurally reverse engineers a binary before answering a user's prompt built in to the tools reverse engineers already use.
Decompiler output
Renaming of unknown variables
Renames all functions, local and global variables and creates and modifies structures.
Follow-up reasoning
Full conversation
After the first response, continue to ask questions related to the binary. The engine will decide whether it needs more context and will either re-reverse the necessary functions or answer based on previous knowledge.
Capabilities
An analysis engine with a conversational layer.
Analysis
Full call tree analysis
Point it at any function and it maps the entire call tree: every callee, every dependency, and the relationships needed to understand the code as a system.
Speed
Parallel by design
Every function in the call tree can be analysed concurrently. A deep graph does not have to wait behind a single agent loop.
Refinement
Bottom-up passes
After analysing a callee, Decompile.re revisits its callers with the new context. Opaque behavior becomes clear once dependencies are known.
Writeback
Auto-rename and retype
Meaningful names are applied directly in the decompiler as analysis runs: functions, locals, and struct members.
Virtual tables
Virtual call resolution
Virtual calls are no longer dead ends. The engine resolves which virtual function belongs to the call and continues reversing.
Chat
Conversational UI
Talk to the tool as if you're talking to a reverse engineer at your desk. Ask follow-up questions, request renames or explanations, and keep drilling into a function without losing analysis context.
Compatibility
Built for the tools reverse-engineers already use.
Decompile.re supports IDA Pro, Ghidra, and Binary Ninja through open source client plugins built for each reverse-engineering environment.
IDA Pro
Open source Python plugin
Ghidra
Open source Java plugin
Binary Ninja
Open source Python plugin
Pricing
Subscription pricing, starting from $10/month.
Each paid tier includes monthly analysis usage. Track only the percentage used and extend it when needed.
Free
$0
Try the analysis engine before choosing a plan.
Usage Limited one-time analysis usage
- Fast model only
- Limited trial usage
- Public-binary CTF demos
- No card required
Plus
$10/mo
For occasional analysis and independent research.
Usage Base monthly analysis allowance
- Fast + Smart models
- Custom usage extensions from $5
- Full product access
- Monthly allowance resets each billing period
Most popular
Boost
$20/mo
For regular reverse-engineering work.
Usage 2x Plus monthly analysis allowance
- Fast + Smart models
- Custom usage extensions from $5
- Full product access
- Monthly allowance resets each billing period
Pro
$50/mo
For professional researchers and daily use.
Usage 5x Plus monthly analysis allowance
- Fast + Smart models
- Custom usage extensions from $5
- Full product access
- Monthly allowance resets each billing period
Elite
$100/mo
For intensive analysis and security teams.
Usage 10x Plus monthly analysis allowance
- Fast + Smart models
- Custom usage extensions from $5
- Full product access
- Monthly allowance resets each billing period