Skip to content

Back to the catalog

linkedin.comarticle

AI Agents for Cybersecurity: A Guided Tour of a Repository with Tons of Resources and Examples

Omar Santos walks through his AI-agents-for-cybersecurity GitHub repo, which has eight parts. Parts 1–4 cover LLM basics (LangChain, prompting, LCEL, RAG). Parts 5–8 cover building security agents with LangGraph, CrewAI, DeepAgents and MCP servers (nmap, CISA KEV, Shodan), and a hardening baseline for six ready-made agentic frameworks. They also cover token-cost control (caching, routing, budget guards, cost per accepted outcome) and a supervisor-led, human-gated security-assessment harness spec with single-use approvals, a kill switch and an append-only audit log. The post also covers Project CodeGuard rules for AI coding assistants. It is a practical learning path and a reference for governing and securing agents.

TL;DR Guided tour of Omar Santos's eight-part AI-agents-for-cybersecurity repo: building SOC and pentest agents, MCP servers, agent hardening, tokenomics, and a human-gated multi-agent assessment spec.
Open original ↗

Organisations named here

  • Artificial Analysis Cyber Index
  • CoSAI
  • MITRE ATT&CK
  • Omar Santos
  • OWASP LLM10:2025

Technologies named here

  • AI-agents-for-cybersecurity
  • LangChain 1.x
  • FastMCP
  • langchain-mcp-adapters
  • CISA Known Exploited Vulnerabilities Catalog
  • NVD API
  • Exploit-DB
  • subfinder and nuclei
  • Chroma
  • Streamlit
  • Hugging Face / DistilBERT
  • Ollama
  • uv
  • TensorFlow Embedding Projector
  • Splunk MCP
  • Part 6 agentic frameworks
  • Agent Skills, Agent Client Protocol, A2A-style messaging
  • Artificial Analysis Cyber Index
  • 07_token_budget_guard.py
  • MITRE ATT&CK
  • LangChain
  • LangGraph
  • CrewAI
  • deepagents
  • Model Context Protocol
  • Nmap
  • Shodan
  • Python
  • Project CodeGuard
  • OWASP Top 10 for Agentic Applications
  • Foundry Security Spec
  • CISA KEV
  • Cisco

Techniques named here

  • ReAct agent loop with security tools
  • Agentic RAG
  • Structured output for reliable routing; LangGraph reducers
  • Attack-surface management pipeline
  • Command injection via crafted hostname
  • Indirect prompt injection
  • Skill and plugin supply-chain attacks
  • Persistent Memory Poisoning
  • Over-broad credentials; exposed control surfaces
  • Approval fatigue with fail-open defaults
  • Denial of wallet
  • Shared cache leakage across tenants
  • Simulated SOC scenario
  • Prompt Injection

Related in AI Security