Security research
Attacker technique, explained for defenders
Explain advisories, write detection rules, review patches and read obfuscated code. Unrestricted models discuss the attacker side typical assistants refuse, within the Acceptable Use Policy and on systems you are authorized to test.
Get API key Read the quickstart
Unrestricted means fewer refusals, not no rules. Read the Acceptable Use Policy
Output a security team can review.
Every example is text for you to check, test and deploy in your own tools.
encoded_powershell.ymlSigmaDetection rules
Describe the behavior and get a Sigma, YARA or Suricata rule to review and tune.
patch-review.mdPatch reviewDoes the fix close the hole?
Paste the advisory and the diff, and ask whether the patch covers the path the advisory describes.
incident-417/loader.jsCode readingObfuscated code, explained
Hand over a script from an incident and get a plain reading of what it does and what to look for.
How a security team uses it.
The model reads and writes. Running, scanning and verifying stay in your environment.
You
Bring the context
The advisory, the source, a log excerpt or decompiled output.
Modellane
Explains and writes one output
A rule, a patch review, a test plan or a report section. Narrow requests give answers you can check.
You
Verify it in your environment
Run rules against your own telemetry and test patches yourself.
Modellane
Keeps no copy
Prompt and reply are discarded when the response ends. Only a metadata record stays.
What we do, and what we don't.
Modellane is a model behind an API, not a security product.
Read the Acceptable Use Policy
What we do
- Explain advisories, exploits and attacker techniques in prose
- Write and review detection rules, patches and test plans
- Read source, decompiled or obfuscated code you paste in
What we don't do
- Scan, probe or connect to any host
- Execute code, rules or samples, or confirm that a fix works
- Store your prompts or the model's replies
- Allow uses the Acceptable Use Policy prohibits, whatever the model answers
Which models fit: Models whose id contains -unrestricted explain exploits, malware behavior and attacker techniques instead of refusing the topic. Standard models keep their base model's behavior, behind the same API. See every model
FAQ
Cybersecurity questions
- Does Modellane scan or test my targets?
- No. The API only reads the text you send and writes text back. It never connects to a host.
- Can I paste malware or decompiled code?
- Yes, as text. Nothing you send is executed, and nothing is stored after the response.
- Which rule formats does it write?
- Any text format, such as Sigma, YARA, Suricata, Snort, KQL or SPL. Validate every rule in your own tooling before you deploy it.
- What is not allowed?
- Creating or deploying malware, or attacking systems you are not authorized to test. The Acceptable Use Policy lists the rest.
Bring your next advisory.
Create a key and send the model the detail a typical assistant would refuse to discuss.