Skip to content
New

DeepSeek V4.1 Flash free for 7 days: 10M tokens a day after a top-up of $10 or more. Until Nov 30.

How it works

Cybersecurity Knowledge Engine 2.0

CKE 2.0 is the security knowledge the model searches before it answers: methodology, payloads, commands and tool docs, disclosed bug bounty reports, a CVE mirror with EPSS and KEV, and a graph that links each CVE to its weakness, the attack patterns that exploit that weakness and the ATT&CK techniques they map to. It is on by default in the chat, the CLI and the editor extension, and each of them can switch it off.

A real path from the graph, starting at a recent CISA KEV entry. Pass any ID in the chain and kb_related returns its neighbors.
  1. CVE

    CVE-2026-115731

  2. CWE

    CWE-94

  3. CAPEC

    CAPEC-35

  4. ATT&CK

    T1027.006

What it holds right now

Counted from the live index every hour. Last count at ⁦05:12 UTC⁩.

Corpus

Documents
110,288
Searchable passages
541,033
Sources
1,404
Graph links
388,951

Vulnerability and bug bounty data

CVE records
408,854
In CISA KEV
1,782
HackerOne reports
15,219
Full report write-ups
12,497

It reads the record before it answers.

Graph path from a recent CISA KEV entry

Path 1 of 5

CVE

CVE-2026-115731Perforce Puppet Enterprise Deserialization of Untrusted Data Vulnerability

CVSS
9.9CRITICAL
EPSS
0.366%percentile 30.9%
CISA KEV
added Nov 17, 2026

What happens before the model answers

  1. 1/4

    It searches

    The question goes to keyword and semantic search at the same time, and the two rankings are merged. One search finds the exact command, flag or ID and also the passages that say the same thing in other words.

    kb_search, kb_read

  2. 2/4

    It pivots

    From a CVE it walks the graph to the weakness behind it, then to the attack patterns that weakness enables and the ATT&CK techniques they map to. Each link is documented by the source that publishes it.

    kb_related

  3. 3/4

    It checks the record

    For each CVE it reads the mirrored record: CVSS score and vector, EPSS score and percentile, KEV status with the date added, and the affected versions where NVD lists them. Package advisories come from OSV.

    cve_lookup, cve_search, advisory_lookup

  4. 4/4

    It cites

    What it used comes back as numbered citations with the source, and the evidence panel opens each one beside the answer.

Coverage by area

Documents per area. A source that covers two areas counts in both.
  • Web applications47,76743%
  • Offensive technique47,04143%
  • Tools and commands42,91239%
  • APIs21,09219%
  • Malware analysis18,43417%
  • Reverse engineering11,60111%
  • ATT&CK techniques10,1589%
  • Incident response6,9826%
  • OSINT & recon6,2456%
  • Cloud6,1186%
  • Game security5,5535%
  • Binary exploitation4,5734%
  • Cryptography3,7703%
  • Mobile1,6121%
  • Digital forensics / DFIR1,0211%
  • Hardware and IoT9591%
  • Systems & networking7811%
  • Social engineering271<1%
  • AI/LLM security238<1%

Where you can use it

With CKE 2.0 on, the chat calls all of it on its own. The CLI, MCP and the Research API reach search and CVE data under your control. The graph is in the chat.

CVE record

Chat
cve_lookup
CLI
notholy cveagent: cve_lookup
MCP
cve_lookup
Research API
GET /v1/cve/{id}

CVE search

Chat
cve_search
CLI
notholy cve search
MCP
cve_search
Research API
GET /v1/cve/search

Package advisories

Chat
advisory_lookup
MCP
advisory_query
Research API
POST /v1/advisories/query

CKE 2.0

Chat
kb_searchkb_read
CLI
notholy kbagent: kb_search
MCP
kb_searchkb_read
Research API
GET /v1/kb/searchGET /v1/kb/{id}

Entity graph

Chat
kb_related

What is inside

Curated corpus

CKE 2.0 is a curated library of methodology, payloads, commands and tool syntax. It is on by default, and while it is on the model searches it on its own. Switch it off in the composer and it stays off in that chat and in the next ones until you switch it back.

Methodology and technique

kb_search, kb_read

Testing methodology and attack techniques the model can cite: OWASP WSTG, ASVS, MASTG and the API Top 10, the Web Security Academy, HackTricks, PayloadsAllTheThings, GTFOBins and LOLBAS. Techniques map to MITRE ATT&CK, CWE and CAPEC, and the library also holds Atomic Red Team tests, the OSSTMM method and a social engineering framework. NIST CSF and CISA advisories cover the defensive side. The model reads this material to explain and plan, and runs none of it.

Commands and tooling

kb_search, kb_read

Everyday command and shell reference for Linux, Windows and PowerShell, with Sysinternals, and networking and traffic tooling: tcpdump, Wireshark, nmap and iproute2. OSINT search operators and dork references. Docs for offensive and defensive tools: sqlmap, ffuf, Burp, Caido, mitmproxy, Nuclei, Metasploit, BloodHound, Certipy and the cloud tools Prowler, Trivy and Stratus Red Team. Game security, reverse engineering and forensics with radare2, x64dbg, Frida, TitanHide and Plaso.

Disclosed bug bounty reports

kb_search, kb_read

Public HackerOne disclosures, classified by weakness, vulnerability class and CWE. Where the full write-up is public, its text is indexed too, so the model can read how the bug was found, proved and fixed, and cite the report.

Vulnerability data

Severity and exploitation data from the sources that publish it, each number with its date. A CVE with no CVSS shows as not scored, and nothing is estimated.

CVE record

cve_lookup

The consolidated record for one CVE: NVD description, CVSS 3.1 or 4.0 score and vector, CWE, affected CPEs with fix versions where NVD has them, references, EPSS score and percentile, CISA KEV status with date added and known ransomware use.

CVE search

cve_search

Filters by text, vendor, KEV status, minimum CVSS and minimum EPSS, then ranks by text match and EPSS. Feed it the products from a scan and the CVEs most likely to be exploited come up first.

Package advisories

advisory_lookup

Known vulnerabilities for a package and version in any OSV ecosystem. Attach a lockfile and ask which pinned dependency is affected.

Ask anything your scope allows.

Create an account, add credits and ask the first question in the chat or the terminal.

  1. 1/3

    Create the account

    Individual or company. You can switch to a company later without losing balance or keys.

    IndividualCompany
  2. 2/3

    Add credits

    By card. Pay per token, no subscription.

    $5.00minimum deposit
  3. 3/3

    Ask

    In the chat, or install the CLI and run notholy login in your terminal.

    $ notholy login