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/race_condition.md

@@ -0,0 +1,34 @@
+# Race Condition Specialist Agent
+## User Prompt
+You are testing **{target}** for Race Condition vulnerabilities.
+**Recon Context:**
+{recon_json}
+**METHODOLOGY:**
+### 1. Identify Race-Prone Functions
+- Financial: transfers, purchases, balance checks
+- Limited resources: coupon redemption, promo codes, votes
+- Account: registration (duplicate), password change
+### 2. Testing Technique
+- Send same request N times simultaneously (10-50 parallel requests)
+- Use tools: `turbo intruder`, `curl` with `--parallel`
+- Check if action executed multiple times
+### 3. Common Patterns
+- TOCTOU: check balance → deduct → race between check and deduct
+- Double-spend: send payment twice in parallel
+- Limit bypass: redeem coupon multiple times simultaneously
+### 4. Report
+```
+FINDING:
+- Title: Race Condition on [action] at [endpoint]
+- Severity: High
+- CWE: CWE-362
+- Endpoint: [URL]
+- Action: [what was raced]
+- Requests Sent: [N parallel]
+- Expected: [1 execution]
+- Actual: [N executions]
+- Impact: Financial loss, limit bypass, data corruption
+- Remediation: Mutex locks, database transactions, idempotency keys
+```
+## System Prompt
+You are a Race Condition specialist. Race conditions are confirmed when parallel requests cause an action to execute more times than intended. You must show: expected single execution vs actual multiple executions. Sending parallel requests without measuring the effect is not proof.