Vulnerability Genome Methodology

1. Why Textual CVE Descriptions Fail

When security teams analyze new CVE disclosures, they encounter three systemic obstacles:

Semantic Ambiguity: Two vulnerabilities labeled “Remote Code Execution” can have completely divergent operational architectures — one requiring complex Java deserialization gadget chains over proprietary binary protocols, while another requires trivial unauthenticated HTTP POST form injection.
Hidden Lineages & Recurrence: Over 40% of critical zero-days in enterprise gateways (VPNs, firewalls, NACs) are direct variants or patch bypasses of flaws discovered years prior. Without phylogenetic tracking, organizations fail to anticipate that a vendor’s bug fix only patched one specific gadget rather than the underlying primitive.
Inability to Predict Exploit Recombination: Threat actors combine known primitives (e.g. prompt injection + Python REPL tool execution) into novel composite weapons.

The Vulnerability Genome solves this by establishing a machine-readable structural standard for exploit mechanics.

2. Interactive Genome Preview

Explore the genetic loci of high-profile vulnerabilities and compare alleles side-by-side:

Vulnerability Genome Sequencer

6-Loci functional decomposition, phylogenetic lineages, and Genetic Distance (Dg)

SEQUENCED VULNERABILITY:

THE 6 STRUCTURAL LOCI SEQUENCE (ALLELES)

Family

MUTATION

Notes...

Phylogenetic Ancestors:

Genetic SHA-256 Fingerprint: ...

3. The 6 Structural Loci

Every vulnerability in Hermes Codex is decomposed across 6 standardized genetic loci:

+-------------------+--------------------+--------------------+--------------------+--------------------+--------------------+
|  1. INGRESS GATE  | 2. ROOT PRIMITIVE  | 3. EXECUTION TIER  |  4. PREREQUISITES  | 5. WEAPON EXPRESS. | 6. RESISTANCE / EV |
|  Network / Auth   |  CWE Core Failure  | OS / Process Level | Activation Signals | Impact Manifest.   | WAF / Evasion Gate |
+-------------------+--------------------+--------------------+--------------------+--------------------+--------------------+

Locus 1: Ingress Gate

The protocol boundary and network interface through which the payload enters the system:

UNAUTH_HTTP_REST: Public unauthenticated REST HTTP endpoints.
BINARY_RPC_T3_IIOP: Proprietary binary RPC listeners (e.g. Oracle T3/IIOP, Java RMI).
PROMPT_CONTEXT_STREAM: Unstructured natural language inputs processed by LLM reasoning loops.
EMAIL_TEMPLATE_PARSE: Server-side template rendering triggers.

Locus 2: Root Primitive (Core CWE)

The foundational software pathology enabling manipulation:

DESERIALIZATION_GADGET (CWE-502)
TEMPLATE_INJECTION_SSTI (CWE-1336)
AUTH_LOGIC_BYPASS (CWE-287)
CONTEXT_BOUNDARY_COLLAPSE (CWE-94)
MEMORY_CORRUPTION_OOB (CWE-787)

Locus 3: Execution Tier

The runtime environment and privilege context where the payload executes:

KERNEL_RING0: Operating system kernel space.
ROOT_SYSTEM_DAEMON: Privileged system services or root-level daemons.
APPLICATION_WORKER: Unprivileged web application worker (e.g. www-data, nobody).
LLM_AGENT_RUNTIME: Autonomous agent decision loop with dynamic tool access.

Locus 4: Receptors & Prerequisites

Environmental conditions required for the gene to express itself:

Specific modules enabled (e.g. Guest Portal, SAML SSO provider).
Network exposure requirements (e.g. direct TCP port connectivity).

Locus 5: Weapon Expression

The observable outcome when the vulnerability is actively weaponized:

ARBITRARY_SHELL_RCE: Direct operating system shell command execution.
TOKEN_REPLAY_IMPERSONATION: Session token forging or administrative impersonation.
PERSISTENT_BACKDOOR_DROP: Installation of persistent web shells or binary implants.
AGENT_SYSTEM_SUBVERSION: Complete takeover of agentic reasoning and tool invocation.

Locus 6: Mutation Resistance & Evasion

Defensive circumvention mechanisms baked into the exploit vector:

Bypasses for Web Application Firewalls (WAF), JEP 290 object filters, or lexical prompt sanitizers.

4. Mathematical Formulation: Genetic Distance ($D_g$)

To evaluate how closely two vulnerabilities share functional exploit architecture, Hermes computes the Genetic Distance ($D_g$) and Similarity Index ($S_g$):

S_g(A, B) = [ sum_{i=1}^{6} w_i * delta(L_i^A, L_i^B) ] / [ sum_{i=1}^{6} w_i ]

D_g(A, B) = 1 - S_g(A, B)

Where:
- L_i represents the allele at locus i.
- delta(L_i^A, L_i^B) in [0, 1] is the matching function between alleles.
- w_i are locus sensitivity weights:
  - w_primitive = 2.5
  - w_weapon = 2.0
  - w_ingress = 1.5
  - w_execution = 1.5
  - w_family = 1.5
  - w_mutation = 1.0

Interactive Workbench: Vulnerability Genome Launch the interactive DNA sequencer and compare arbitrary CVE genomes side-by-side.

Methodology: Threat Trajectory (HTR) Explore how risk velocity and acceleration correlate with phylogenetic mutation types.