Skip to content

Add nation-state level security assessment module with PoC and risk scope#84

Open
kingassune wants to merge 4 commits intoConway-Research:mainfrom
kingassune:main
Open

Add nation-state level security assessment module with PoC and risk scope#84
kingassune wants to merge 4 commits intoConway-Research:mainfrom
kingassune:main

Conversation

@kingassune
Copy link

@kingassune kingassune commented Feb 19, 2026

Adds a security assessment module that evaluates the automaton's defenses against APT-level threats across 10 attack vectors, with live proof-of-concept testing and CVSS-aligned risk scoring.

New module: src/security/

  • types.ts — Threat actors, attack vectors, severity levels, findings, PoC, and risk scope type definitions
  • threat-vectors.ts — 10 vector evaluators (prompt injection, supply chain, cryptographic, identity/auth, persistence tampering, lateral movement, exfiltration, financial manipulation, self-modification abuse, social engineering). Three vectors run live payloads against the injection defense pipeline; others perform static analysis
  • poc.ts — PoC generators that exercise sanitizeInput() with multi-stage attack payloads (boundary escape, authority spoofing, obfuscation chaining)
  • risk-scope.ts — Computes impact across CIA + financial + reputational dimensions, identifies cascading risks (e.g. supply chain → child agent propagation), and measures attack surface metrics
  • assessment.ts — Orchestrates evaluation and produces a formatted report

Usage

import { runSecurityAssessment, formatAssessmentReport } from './security/assessment.js';

const assessment = runSecurityAssessment(); // all 10 vectors, nation_state threat model
const report = formatAssessmentReport(assessment);
// assessment.overallRiskScore → 6.6/10.0 (MEDIUM)
// assessment.findings → 4 mitigated, 5 partially mitigated, 1 vulnerable

Key findings from the assessment

Finding CVSS Status
Multi-stage prompt injection (base64 obfuscation gap) 9.8 Vulnerable
Supply chain / skill loader unsigned code 7.5 Partial
Data exfiltration via curl (not in forbidden patterns) 7.8 Partial
Self-modification safety guardrails 5.0 Mitigated
Financial asset protection 2.5 Mitigated

All findings are MITRE ATT&CK mapped. 21 new tests; all 31 tests pass. CodeQL clean.

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

  • mainnet.base.org
    • Triggering command: /opt/hostedtoolcache/node/24.13.0/x64/bin/node /opt/hostedtoolcache/node/24.13.0/x64/bin/node --conditions development --conditions node /home/REDACTED/work/automaton/automaton/node_modules/tinypool/dist/entry/process.js (dns block)
    • Triggering command: /opt/hostedtoolcache/node/24.13.0/x64/bin/node /opt/hostedtoolcache/node/24.13.0/x64/bin/node --conditions development --conditions node /home/REDACTED/work/automaton/automaton/node_modules/tinypool/dist/entry/process.js 13.0/include/node k/_temp/ghcca-node/node/bin/git 13.0/src (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

Original prompt

Run a nation state level security assessment with poc and risk scope

Copilot AI and others added 4 commits February 19, 2026 02:18
…cope

Co-authored-by: kingassune <6126851+kingassune@users.noreply.github.com>
…r documentation, remove accidental package-lock.json

Co-authored-by: kingassune <6126851+kingassune@users.noreply.github.com>
…-scope

Add nation-state security assessment framework with PoC and risk scoping
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants