r2ai — LLM assistance for radare2
For legal, authorized and educational use only. Only analyze software you own or have written permission to examine.
Quick answer
r2ai (radareorg/r2ai) brings LLM-assisted reversing to the radare2 world. Release 1.5.3 (2026-10-18 UTC) is MIT-licensed. Use it after you can already navigate radare2 on binaries you are allowed to analyze.
Snapshot
| Field | Value |
|---|---|
| Ecosystem | radare2 |
| License | MIT |
| Latest checked | 1.5.3 (2026-10-18 UTC) |
| Stars (order of magnitude) | ~0.5k on GitHub (counts change daily) |
| Job | LLM-assisted RE on top of r2 |
Fit vs neighbors
| If you already use… | Nearby AI helper |
|---|---|
| Ghidra GUI | GhidrAssist |
| Ghidra + agents | GhidraMCP / REA |
| radare2 CLI | r2ai |
| Rizin + Cutter | Stay with Cutter scripting; see rizin vs radare2 |
No ranking — choose the stack your team can support.
Legal boundaries
- Authorized / educational targets only.
- Do not paste customer secrets into public model endpoints.
- Mentions of crack/keygen appear only as refusals.
Which to pick
- r2ai amplifies radare2; it does not replace learning
aaa/ seek / print. - Keep model traffic inside your compliance boundary.
- Prefer official radareorg builds.
For legal, authorized and educational use only.
FAQ
What is r2ai?
An open-source project for LLM-based reversing workflows integrated with radare2.
Which version is cited?
1.5.3 published 2026-10-18 UTC (checked 2026-10-24).
License?
MIT per GitHub SPDX on radareorg/r2ai.
Do I still need radare2?
Yes. r2ai sits on the radare2 ecosystem; install and learn r2 first for authorized targets.
Can r2ai bypass DRM?
No. Topic plan §4 excludes circumvention; BinBench will not document it.