Agent Firewall
БесплатноНе проверенPanovista provides military-grade, FIPS 140-3 compliant L7 security proxies for Enterprise AI deployments. Our Zero-Trust architectural boundaries and offline c
Описание
Panovista provides military-grade, FIPS 140-3 compliant L7 security proxies for Enterprise AI deployments. Our Zero-Trust architectural boundaries and offline cryptographic licensing ensure your AI agents operate securely within strict PCI-DSS and Article 12 compliance frameworks.
README
Docker Pulls Release Security Policy
Panovista L7 Agent Firewall (Evaluation Edition) - Core V3.0
🌐 Visit panovista.io | Zero-Trust, Stateless L7 Security Proxy for Enterprise AI Deployments.
Panovista provides an offline, cryptographically locked architectural boundary for AI Agents and Model Context Protocol (MCP) servers. By dropping this stateless sidecar proxy into your network, you instantly enforce strict L7 routing constraints, payload validation, and compliance logging without a single external ping to an outside vendor server.
The Frictionless 14-Day Evaluation
This public repository provides access to the 14-Day Evaluation Tier of the V3.0 Core Engine.
Unlike older versions, this evaluation container does not require an upfront token to boot. Instead, it uses a Frictionless First-Boot mechanism. The moment the container launches on your network, it securely creates an internal cryptographic timestamp file within your mounted state volume. The proxy will seamlessly route traffic for 14 calendar days. On day 15, the evaluation window closes, and the proxy will permanently lock itself until a valid 12-month enterprise license token is provided.
State Persistence: Panovista utilizes a local volume at /var/lib/panovista to track the evaluation trial period. Ensure this directory is mapped to a persistent host volume in your docker-compose configuration. Deleting this directory or failing to map it will reset your evaluation window.
Quick Start Deployment
Deploying the Panovista proxy requires zero external dependencies. It runs completely offline in your local environment with a strictly bounded memory footprint.
1. Prepare Your Host Directories
Create an empty local directory named state to house the persistent evaluation trial clock metadata:
mkdir -p state
2. Create Your docker-compose.yml
Copy the following streamlined configuration file into your target environment folder.
version: '3.8'
services:
panovista-security-proxy:
image: ghcr.io/panovista/agent-firewall:latest
container_name: panovista-core-v3
ports:
- "8080:8080"
volumes:
# Mandatory persistent directory for tracking the 14-day evaluation clock
- ./state:/var/lib/panovista # Ensures the trial.lock file persists across restarts
- ./panovista-config:/etc/panovista/policies:ro
environment:
- PANOVISTA_PORT=8080
- TARGET_MCP_URL=http://your-internal-mcp-server:80
- PANOVISTA_ENV_ID=eval-vpc-local
- PANOVISTA_ALLOW_HEADLESS=TRUE
# ==========================================================
# 14-DAY TRIAL ACTIVE BY DEFAULT
# To upgrade to the 12-Month Paid Tier, uncomment below:
# - PANOVISTA_LICENSE_TOKEN=pv_lic_YOUR_TOKEN_HERE
# ==========================================================
3. Boot the Firewall
Run the following command to pull the signed Panovista image and boot the engine:
docker compose up -d
4. Verify Telemetry & Trial Status
Check your container logs to ensure the evaluation clock initialized, the persistent storage mounted, and the strict L7 traffic parsing is active:
docker compose logs panovista-security-proxy
You should see our structured Passive Telemetry metrics outputting directly to stdout, confirming a successful local first boot:
panovista-core-v3 | 2026/07/10 12:19:25 [*] No license token provided. Initiating Evaluation Tier checks...
panovista-core-v3 | 2026/07/10 12:19:25 ⏱️ FIRST BOOT DETECTED. 14-Day Evaluation Clock Started.
panovista-core-v3 | 2026/07/10 12:19:25 ⚠️ EVALUATION MODE ACTIVE. 14 Days Remaining.
panovista-core-v3 | 2026/07/10 12:19:25 {"level":"info","tag":"PANOVISTA_METRIC","status":"boot","node_id":"8ea012c93647","license_tier_claimed":"core_v3","uptime_seconds":0,"peak_concurrent_streams":0}
panovista-core-v3 | 2026/07/10 12:19:25 🚀 Panovista Core V3 active on port 8080
Verify the container readiness for your orchestrator via our native health probes:
# Returns 200 OK instantly if the HTTP server socket loop is up
curl http://localhost:8080/health/live
# Returns 200 OK only after trial validation passes or license verifies
curl http://localhost:8080/health/ready
Configuration Lexicon (V3.0 Core)
Panovista's runtime behavior is controlled cleanly via environment variables and local volume mappings.
| Variable Name | Required | Default | Functional Description |
|---|---|---|---|
TARGET_MCP_URL |
Yes | None | The internal network URL of the raw upstream tool server or LLM backend being shielded. |
PANOVISTA_ENV_ID |
Yes | None | Ties binary execution to a specific VPC or environment ID for isolation matching. |
PANOVISTA_ALLOW_HEADLESS |
Yes | None | Bypasses the anti-automation security lock for cloud orchestration engines (TRUE). |
PANOVISTA_PORT |
No | 8080 |
Local port the proxy core listens on for incoming AI agent or client application requests. |
PANOVISTA_LICENSE_TOKEN |
No | None | The offline, cryptographically signed Ed25519 token string that unlocks the unlimited tier. |
Instant Attack Deflection Test
Prove Panovista is actively shielding your backend parameters. Run a standard curl command attempting a Layer 4 Server-Side Request Forgery (SSRF) bypass through the proxy:
curl -X POST http://localhost:8080/ \
-H "Content-Type: application/json" \
-d '{
"jsonrpc": "2.0",
"id": "verification-test-001",
"method": "chat",
"params": {
"name": "agent",
"arguments": {
"target": "http://169.254.169.254/latest/meta-data/"
}
}
}'
Expected Defensive Response
The proxy instantly intercepts the forbidden cloud infrastructure address, terminates the transaction safely before it breaches your perimeter, and returns a sanitized JSON-RPC error payload:
{
"jsonrpc": "2.0",
"id": "verification-test-001",
"error": {
"code": -32001,
"message": "Security Policy Violation"
}
}
Upgrading to Sovereign Enterprise
The Panovista Sovereign Enterprise tier transitions your infrastructure to our strictly-versioned private registry, removes the 14-day trial clock file limitations, and unlocks unlimited enterprise scale.
Sovereign builds include:
- Native compilation using a FIPS 140-3 validated cryptographic toolchain.
- Hardcoded defensive thresholds: 5MB egress volume choking, 300 req/min rate limiting, and a 20-layer deep JSON recursion filter.
- Injectable PEM-encoded Public Key certificates (
JWT_PUBLIC_KEY) for full IdP identity access verification.
To upgrade your evaluation environment to a permanent enterprise license, visit panovista.io or contact our sales engineering team at [email protected] to receive your scoped Cryptographic Enterprise Token.
Установка Agent Firewall
У этого сервера нет опубликованного пакета — он собирается из исходников. Открой репозиторий и следуй инструкции в README.
▸ github.com/panovista/agent-firewallFAQ
Agent Firewall MCP бесплатный?
Да, Agent Firewall MCP бесплатный — установка в пару кликов через Unyly без оплаты.
Нужен ли API-ключ для Agent Firewall?
Нет, Agent Firewall работает без API-ключей и переменных окружения.
Agent Firewall — hosted или self-hosted?
Self-hosted: сервер запускается локально на твоей машине командой из раздела установки.
Как установить Agent Firewall в Claude Desktop, Claude Code или Cursor?
Открой Agent Firewall на unyly.org, выбери вкладку своего клиента (Claude Desktop, Claude Code, Cursor) и нажми Install — конфиг сгенерируется автоматически, без правки JSON.
Похожие MCP
Fetch
Web content fetching and conversion for efficient LLM usage.
AWS KB Retrieval
Retrieval from AWS Knowledge Base using Bedrock Agent Runtime.
автор: modelcontextprotocolSpring AI MCP Server
Provides auto-configuration for setting up an MCP server in Spring Boot applications.
llm-analysis-assistant
A very streamlined mcp client that supports calling and monitoring stdio/sse/streamableHttp, and can also view request responses through the /logs page. It also
автор: xuzexin-hzMCP-Agent
A simple, composable framework to build agents using Model Context Protocol by [LastMile AI](https://www.lastmileai.dev)
автор: lastmile-aiSpring AI MCP Client
Provides auto-configuration for MCP client functionality in Spring Boot applications.
mcp.natoma.ai
A Hosted MCP Platform to discover, install, manage and deploy MCP servers by [Natoma Labs](https://www.natoma.ai)
MCPHub
Website to list high quality MCP servers and reviews by real users. Also provide online chatbot for popular LLM models with MCP server support.
MCP Servers Rating and User Reviews
Website to rate MCP servers, write authentic user reviews, and [search engine for agent & mcp](http://www.deepnlp.org/search/agent)
mkinf
An Open Source registry of hosted MCP Servers to accelerate AI agent workflows.
Compare Agent Firewall with
Не уверен что выбрать?
Найди свой стек за 60 секунд
Автор?
Embed-бейдж для README
Похожее
Все в категории ai
