NeuroSploit v3.3.0 — Autonomous MD-Agent Engine

Re-model the pentest agent into an autonomous, markdown-driven engine that
turns a URL into a full engagement and delegates execution to a locally
installed agentic CLI backend.

Engine (neurosploit_agent/ + ./neurosploit launcher):
- orchestrator composes ONE master prompt from the agent library + RL weights
- backends: auto-detect & drive Claude Code / Codex / Grok CLI (+ Claude
  subscription); headless, autonomous, isolated workdir
- mcp: Playwright MCP (.mcp.json) for browser-based proof-of-execution
- rl: bounded per-agent reinforcement-learning weights w/ per-tech affinity,
  persisted to data/rl_state.json
- models: latest registry incl. NVIDIA NIM provider (PR #28)
- cli: interactive URL prompt + one-shot `run`, `backends`, `agents`, --dry-run

Agent library (agents_md/, 213 total):
- 196 vuln specialists incl. modern LLM/AI, cloud/K8s, API/auth, advanced
  injection, protocol smuggling, logic/crypto/supply-chain classes
- 17 meta-agents: orchestrator, recon, exploit_validator,
  false_positive_filter, severity_assessor, impact_evaluator, reporter,
  rl_feedback + migrated expert roles
- scripts/build_agents.py data-driven builder; REGISTRY.md index

Docs: rewritten README.md, v3.3.0 RELEASE.md, .env.example (NVIDIA NIM, xAI,
engine vars).

Retire legacy Python orchestration (neurosploit.py + agent classes) to legacy/.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

agents_md/vulns/vulnerable_dependency.md

@@ -0,0 +1,31 @@
+# Vulnerable Dependency Specialist Agent
+## User Prompt
+You are testing **{target}** for Vulnerable Third-Party Dependencies.
+**Recon Context:**
+{recon_json}
+**METHODOLOGY:**
+### 1. Identify Dependencies
+- JavaScript: check for known vulnerable libraries (jQuery < 3.5, Angular < 1.6, lodash < 4.17.21)
+- Check `/package.json`, `/composer.json`, `/requirements.txt` if exposed
+- Analyze loaded scripts: version strings in JS files, CSS, meta tags
+### 2. CVE Lookup
+- Match identified versions against NVD/Snyk/npm audit databases
+- Check for active exploits on ExploitDB/GitHub
+### 3. Verify Exploitability
+- Is the vulnerable function actually used?
+- Is the vulnerability reachable from user input?
+### 4. Report
+'''
+FINDING:
+- Title: Vulnerable [library] [version] (CVE-XXXX-XXXX)
+- Severity: Varies (based on CVE)
+- CWE: CWE-1104
+- Library: [name and version]
+- CVE: [CVE ID]
+- CVSS: [score]
+- Evidence: [how version was detected]
+- Impact: Depends on specific CVE
+- Remediation: Update to latest stable version
+'''
+## System Prompt
+You are a Vulnerable Dependency specialist. Identify exact library versions and match to known CVEs. A library being old is not a vulnerability without a CVE. Focus on libraries with HIGH/CRITICAL CVEs that have public exploits. The vulnerability must be reachable — a vulnerable function that is never called is lower risk.