Analysis
Reverse engineer binary: summarize behavior, key functions, suspicious/vulnerable items
- Analysis:
- Matched binary to vendor release, extracted strings and imports from a readable copy, tried Ghidra but used metadata + string/import analysis.
- Conclusion:
- Confirmed FrekansLoL v3.0.6 skin-changer: injects DLL into League process, automates LCU actions, performs device fingerprinting and paid license checks; no exfi
Tools: todo_write, run_sandbox, inspect_binary, dump_data, list_files, find_files, web_search, web_fetch, task, read_file, write_file
Run it yourselfReverse engineer this binary: summarize what it does, key functions, and any suspicious or vulnerable behavior.
I'll start by inspecting the binary, then load the Ghidra analysis workflow for deeper analysis.
The file was saved with a + instead of a space (FrekansLoL+v4.exe). Continuing with that name:
Odd — I/O errors reading the file. Let me investigate:
The file sits on a FUSE-backed mount that's returning EIO on reads. Let me probe whether any part is readable and try alternate access paths:
Every read from the shell fails with EIO — the file's contents aren't accessible via the workspace mount. Let me try the analysis tools, which may access the file through a different backend path (/sandbox/):
The analysis tools can read the file (they use a different backend path than the sandbox shell, which is hitting EIO). Good triage info already: 43.7 MB PE32+ x64 Windows GUI app, MSVC-linked, PDB name frekanslol.pdb, importing process-memory APIs (OpenProcess, ReadProcessMemory), DeviceIoControl, networking (ws2_32), crypto (bcrypt, crypt32). Let me run the full Ghidra decompilation — it's a large binary, so this will take a while:
Error calling tool 'dump_data': Ghidra pod did not finish before timeout. Pod log (last 20 lines): Exception while decompiling 1404f89b0: process: timeout Exception while decompiling 140500760: process: timeout Exception while decompiling 140505140: process: timeout Progress: 86197/111149 (77%) Analysis failed for 0x1405107f0 Exception while decompiling 140e2d990: process: timeout Progress: 91502/111149 (82%) Analysis failed for 0x140c54790 Progress: 93405/111149 (84%) Exception while decompiling 140575d80: process: timeout Exception while decompiling 1405852a0: process: timeout Low-level Error: Backward normalization not implemented Progress: 94716/111149 (85%)
Ghidra timed out partway (111k functions — heavily statically-linked binary). Let me check if partial results were written, and retry with what's available:
Error calling tool 'run_sandbox': Could not reach the sandbox to mount the Workspace at /data/google-oauth2-117202453629727959942/1cd85c1e-540b-4077-b8a7-f6ff11861fdf — 1 attempts over 126s all failed (Failed to execute command: ). The command was not run, so retrying it is safe.
Error calling tool 'run_sandbox': Failed to create sandbox: <html> <head><title>504 Gateway Time-out</title></head> <body> <center><h1>504 Gateway Time-out</h1></center> </body> </html>
Error calling tool 'run_sandbox': Failed to create sandbox: <html> <head><title>504 Gateway Time-out</title></head> <body> <center><h1>504 Gateway Time-out</h1></center> </body> </html>
The sandbox is flaky, but the analysis-tool pod is separate. Let me check for partial Ghidra output via the file tools:
No partial output was saved. The tool pod reads the file fine (inspect_binary worked), so let me retry the Ghidra dump, and in parallel search the web to identify what "FrekansLoL" is:
Error calling tool 'dump_data': Ghidra pod finished with phase Failed. Pod log (last 20 lines): Copying binary to local /tmp/FrekansLoL v4.exe... Opening project FrekansLoL v4.exe-569efc... Loading binary FrekansLoL v4.exe... Running Ghidra auto-analysis (this can take minutes)... WARNING: A terminally deprecated method in sun.misc.Unsafe has been called WARNING: sun.misc.Unsafe::staticFieldOffset has been called by org.apache.felix.framework.util.SecureAction (file:/opt/ghidra/Ghidra/Features/Base/lib/org.apache.felix.framework-7.0.5.jar) WARNING: Please consider reporting this to the maintainers of class org.apache.felix.framework.util.SecureAction WARNING: sun.misc.Unsafe::staticFieldOffset will be removed in a future release
Identified via web: FrekansLoL / "Frekans Studio" is a Turkish League of Legends custom skin manager / skin changer (GitHub frekans911/frekanslol), part of the CSLOL/LTK skin-mod ecosystem. Now let me get the source repo details and retry the Ghidra dump in parallel:
The repo confirms: closed-source Windows app, skin changer + mod manager, with in-app account login and paid license activation (explains crypto/network APIs). Sandbox output was empty — let me verify shell health and try radare2 directly (the tool pod can read the file; maybe the shell can too):
Shell access to the file is definitively broken (EIO) — only the analysis pod can read it. Retrying the Ghidra dump (nondeterministic timeouts so far), plus researching whether this binary has been flagged anywhere:
Error calling tool 'dump_data': Ghidra pod finished with phase Failed. Pod log (last 20 lines): Copying binary to local /tmp/FrekansLoL v4.exe... Opening project FrekansLoL v4.exe-569efc... Loading binary FrekansLoL v4.exe... Running Ghidra auto-analysis (this can take minutes)... WARNING: A terminally deprecated method in sun.misc.Unsafe has been called WARNING: sun.misc.Unsafe::staticFieldOffset has been called by org.apache.felix.framework.util.SecureAction (file:/opt/ghidra/Ghidra/Features/Base/lib/org.apache.felix.framework-7.0.5.jar) WARNING: Please consider reporting this to the maintainers of class org.apache.felix.framework.util.SecureAction WARNING: sun.misc.Unsafe::staticFieldOffset will be removed in a future release
Note: the second "Frekans Studio" (frekans.studio, binaural sound lab) is a different project — our binary matches the GitHub frekans911/frekanslol LoL skin changer. Ghidra failed again (43 MB / ~111k functions is beyond the pod's limits). One more attempt, plus checking the app's releases page:
Error calling tool 'dump_data': Ghidra pod finished with phase Failed. Pod log (last 20 lines): Copying binary to local /tmp/FrekansLoL v4.exe... Opening project FrekansLoL v4.exe-569efc... Loading binary FrekansLoL v4.exe... Running Ghidra auto-analysis (this can take minutes)... WARNING: A terminally deprecated method in sun.misc.Unsafe has been called WARNING: sun.misc.Unsafe::staticFieldOffset has been called by org.apache.felix.framework.util.SecureAction (file:/opt/ghidra/Ghidra/Features/Base/lib/org.apache.felix.framework-7.0.5.jar) WARNING: Please consider reporting this to the maintainers of class org.apache.felix.framework.util.SecureAction WARNING: sun.misc.Unsafe::staticFieldOffset will be removed in a future release
Ghidra is definitively unable to finish this binary (4 failures, all in auto-analysis of ~111k functions). The compile timestamp (Nov 2, 02:52 UTC) matches the v3.0.6 release (Nov 2, 03:08 UTC) from the vendor's GitHub. Let me try one more extraction angle — a subagent on the analysis pod (where the file IS readable) to pull strings/components:
Error: FrekansLoL v4.exe is 43789312 bytes, over the 10485760-byte limit; use run_sandbox
The subagent recovered strong evidence — it matched our binary byte-for-byte to the payload inside the vendor's official FrekansStudio_3.0.6_x64-setup.exe release and extracted strings from a readable copy. Let me verify these artifacts and spot-check key claims myself before reporting:
Artifacts are present. Let me verify hashes and spot-check the key string claims independently:
The vendor's installer hash matches its published SHA-256 exactly (71bc2c9a…733 ✔ authentic). Notably, the extracted 43 MB payload also hits EIO from the shell — same storage issue as the original — but the subagent read it via the tool pod. Let me verify the string artifacts and cross-check the payload identity myself:
Identity confirmed: the extracted payload is byte-for-byte the same binary as our file (same size, PDB GUID E6BF1FD3…, link time, computed checksum 0x029C8408, entrypoint, section layout) — and it comes from the vendor's hash-verified official installer. Now let me independently spot-check the subagent's key string claims:
All claims verified. A few more greps to complete the behavioral picture and rule out classic malware indicators:
Module map confirmed. Final negative checks for classic malware behaviors:
Clean negative results. Let me check the two positive hits ("keylog", "clipboard") for context, then finalize:
"NoKeyLog" is a rustls TLS-library internal — benign. All verification complete. Let me finalize the report:
Full report saved to FREKANSLOL_REPORT.md. Summary:
What it is
FrekansLoL / "Frekans Studio" v3.0.6 — a freemium League of Legends skin changer / mod manager by Turkish dev frekans911 (closed-source; GitHub hosts releases only). 43.7 MB unsigned PE32+ x64, MSVC-linked Rust/Tauri 2.10.3 app (WebView2 frontend, tokio/reqwest/rustls). I proved identity: it's byte-identical (PDB GUID E6BF1FD3…, link time 2026‑11‑02 02:52 UTC, checksum 0x029C8408) to the payload of the vendor's official installer, which I hash-verified against the SHA-256 published in the vendor's release notes. Link time is 16 minutes before that release was published.
What it does (key functions, from 22k extracted strings + 475 imports)
- Skin patch engine (
patcher/host.rs): builds WAD patches (vendorltk_*crates), spawnsltk_patcher_host.exeand injectscslol-dll.dllintoLeague of Legends.exevia a sha256/export-verified host↔DLL handshake (cslol_hook); hard kill switchCSLOL_EOL_TIMESTAMP=1794988800(2026‑11‑17). - LCU client automation: harvests
LeagueClientUx.execommand line via PowerShell to get the port +remoting-auth-token, then auto-accepts queue, auto-picks/bans, bench-swaps, honors, replays queues, writes rune pages. - Marketplace/providers:
api.divineskins.gg,runeforge.dev, LeagueSkins repo, Riot CDNs; Pengu Loader install; mod library with.fantome/.modpkgpackaging. - Licensing: paid membership (30 TRY/45 days) with device fingerprinting (
MachineGuid+ SMBIOS UUID + CPU ID →frekanslol-device-v1), minisign-signed updater, GitHub update checks.
Suspicious / risky
- DLL injection into a game process — core mechanism, not malware per se, but a Riot-ToR violation with ban risk. No Vanguard tampering, refuses elevated games, checks IFEO debugger hijacks.
- Credential-access pattern: reads the LCU session token from a process command line (mitigated: scrubs tokens/emails/PUIDs from logs).
- Device fingerprinting + entitlement calls to a closed third-party backend (
divineskins.gg), unsigned distribution, and advice to add itself to antivirus exclusions.
Clean
Exhaustive negative searches found no browser credential theft, miner, clipboard hijack, scheduled tasks, service/driver installs, or file exfiltration. Persistence = optional autostart plugin only.
Caveats
Your sandbox couldn't read the file directly (FUSE EIO — I documented the workaround), and Ghidra couldn't decompile a 43 MB/111k-function binary in 4 attempts, so findings are metadata + full string evidence on a verified identical copy, not function-level decompilation. No dynamic execution was run.
