⚙️ Recommended runtime settings — gemma-native sampling temperature 1.0, top_k 64, top_p 0.95, min_p 0.01 (the min_p 0.01 floor prevents the reasoning-loop empty-answer issue), context length ≥ 16k (32k recommended), and a generous max_tokens when running with thinking on. The Gemma-4 thinking path needs --jinja.

gemma-4-26B-A4B-offsec-expert (GGUF)

Base gemma-4-26B-A4B-it with an Offensive Security & Reverse Engineering FactBank baked into its chat-template. The model answers correctly about post-cutoff / breaking-change APIs in 17 offensive-security / RE libraries — not by fine-tuning, but by carrying a searchable bank of 489 landmine facts that fires inside llama.cpp at inference time. Weights are untouched (only the GGUF chat-template was rewritten); no external RAG service.

The model supplies the reasoning; the bank supplies the knowledge it was never trained on.

🔗 Full project — all experts, methodology, per-question transcripts, and benchmarks: github.com/mhndayesh/experts-models

What this expert is about

Red-team tooling and binary analysis move fast and break APIs constantly — a v2→v3 rewrite (volatility3), a Rust rewrite (yara-x), a tool renamed outright (CrackMapExec → netexec). A stock model answers about these from stale training and is confidently wrong. This bank carries the exact breaking-change facts so the model gets them right.

Everything here is a landmine fact: post-cutoff, reverses-a-trained-habit, or a silent failure the model wouldn't otherwise catch. Every fact is quote-grounded to a real migration guide / changelog line. Examples the base gets wrong and this model gets right: the CrackMapExec→netexec (nxc) rename, the volatility3 v3 plugin shape (PluginInterface+TreeGrid+run), yara-x's Rust rule/API differences, capstone v6, frida 17.

Libraries in the bank (17) — 489 facts total

library	facts	what it is / the churn
`angr`	103	binary analysis / symbolic execution — API migration
`ldap3`	63	LDAP client (AD tooling) — async module rename, constant renames
`capstone`	58	disassembly framework — the v6 API break
`netexec`	52	network execution toolkit — the CrackMapExec successor (`nxc`)
`pwntools`	45	CTF / exploit-dev toolkit — v5 changes
`nuclei`	27	template-based vulnerability scanner — v3 flags/schema
`responder`	27	LLMNR/NBT-NS/mDNS poisoner — option changes
`impacket`	18	network-protocol toolkit — example/API changes
`volatility3`	18	memory forensics — the v2→v3 rewrite (PluginInterface/TreeGrid/run)
`frida`	16	dynamic instrumentation — the v17 split
`unicorn`	15	CPU emulator — the v2 API
`plaso`	12	timeline forensics — tool/flag changes
`certipy`	10	AD CS abuse — v5 command surface
`yara-x`	10	YARA rewritten in Rust — rule/API differences
`dnfile`	6	.NET PE parsing — API changes
`bloodhound-py`	5	BloodHound Python ingestor — CE changes
`pefile`	4	PE-file parsing — API changes

Where the facts come from (mined sources)

Each library's facts were extracted from its migration guide / changelog (source targeting is the whole game — a migration guide, not release-note noise), then quote-verified against the source line:

angr migration guide
ldap3 changelog
capstone v6 changelog
netexec migration notes
pwntools 5 changelog
nuclei 3 changelog
responder changelog
impacket changelog
volatility3 migration guide
frida 17 notes
unicorn 2 changelog
plaso changelog
certipy v5 notes
yara-x differences doc
dnfile changelog
bloodhound-py docs
pefile changelog

Full provenance (the mined source docs themselves) lives in the repo under v2/extractor/experts/offensive-security-re/sources/.

Results (this model — hand-verified)

Same 44 adversarial landmine questions, base vs. this baked model, identical prompts (the bank injects in-engine). Config: thinking-on + authority framing, Gemma-native sampling.

	base 26B-A4B	this model	Δ	error-closure*
/44 landmine questions	17/44 (38.6%)	43/44 (97.7%)	+26	96%

* of the answers the base got wrong, the fraction the bank fixed. Every answer was hand-scored — an automated substring check miscounts in both directions (it fails a correct answer that names the old API as a contrast, and passes a semantically wrong one).

How to run

The bank lives in the chat-template, so retrieval needs the template applied — run on llama.cpp:

llama-server -m gemma-4-26B-A4B-offsec-expert-Q4_0.gguf --jinja --port 8080 --ctx-size 8192

Then query normally — the same prompt you'd send the base model; the bank fires automatically for covered topics. Sampling — use Gemma-native, not a bare low temperature: temperature 1.0, top_k 64, top_p 0.95, min_p 0.01 (the min_p floor prevents reasoning-loop empty answers).

Best accuracy (authority + thinking) — this is how the numbers above were measured. Send chat_template_kwargs={"enable_thinking": true} and a system prompt telling the model the looked-up facts are verified and supersede its training. A reasoning model otherwise tends to "correct" an injected fact back to its trained prior; authority framing holds the fact.

Limitations

Scoped to the 17 covered libraries. Outside them it's the base model.
Supplies knowledge, not reasoning. A multi-step transform can still fail even with the right fact retrieved.
Retrieval gate is token-based, with aliases. This bake includes the gate-alias fix — a rename's OLD name (e.g. "CrackMapExec") also opens the tab — but a wholly unrelated phrasing may still miss.
Numbers are hand-scored landmine tests, not a general coding benchmark.

Papers

The write-ups behind this project (PDFs, rendered on GitHub): Research Report · Idea · Technical · Verdict · Evidence Ledger

Provenance & license

Base: lmstudio-community/gemma-4-26B-A4B-it-QAT-GGUF (Q4_0). This model = that GGUF with tokenizer.chat_template rewritten to embed an inverted-index retriever + the bank (factbank.version 0.4.0, gate-alias fix applied).
License: Google Gemma Terms of Use (license: gemma) — a gemma-4 derivative. The fact bank is from the FactBank project (see the repo LICENSE); mined sources keep their own licenses.
Source & method: github.com/mhndayesh/experts-models.

mhndayesh/gemma-4-26B-A4B-offsec-expert-GGUF