Automated Vulnerability Discovery and PoC Generation

You are a vulnerability research lead and AI security engineer. Guide me through building an automated vulnerability discovery and PoC generation platform: 1. Discovery Pipeline: Combine static analysis (SAST for [INSERT_LANGUAGES]), fuzzing (AFL++, libFuzzer), dependency scanning (SCA), and container/image scanning. Define how AI agents prioritize targets based on exploitability, exposure, and business impact. 2. AI-Assisted Triage: Use an LLM to classify discovered issues, map them to CWE/CVE, assess exploitability, and discard false positives by reasoning about code context and data flow. 3. PoC Generation: Build an agent that generates proof-of-concept exploits for confirmed vulnerabilities. The agent reads the vulnerable code, understands the flaw, crafts a minimal PoC, and validates it against a sandboxed instance. Ensure the PoC is safe (non-destructive) and includes clear reproduction steps. 4. Sandbox & Validation: Design isolated environments (container pools, VM snapshots) where PoCs run safely. Define success criteria (crash, code execution, data exposure) and automatic evidence capture. 5. Responsible Disclosure Workflow: Generate standardized vulnerability reports (CVSS, CWE, affected versions, remediation) and route to development teams via [INSERT_BUG_TRACKER]. Track fix verification and retest.