# PipeLab / Pipelock

> Pipelock is an open-source agent firewall for MCP and AI agent egress. Version 3.3.0 adds an operator dashboard, free evidence viewer, stricter receipt verification, and public benchmark-corpus refresh work while enforcing MCP, HTTP, and WebSocket egress at the network boundary.

## Full context

- [Full site context for agents](https://pipelab.org/llms-full.txt)

## Primary pages

- [Product](https://pipelab.org/pipelock/): Pipelock overview, architecture, transports, threat model, install paths.
- [Proof](https://pipelab.org/proof/): Public Agent Egress Bench corpus, signed rule bundles, and signed assessment evidence.
- [Pricing](https://pipelab.org/pricing/): Community detection is free. Paid tiers add multi-agent coordination, Assess reports, and Enterprise fleet governance.
- [Enterprise](https://pipelab.org/enterprise/): Enterprise fleet governance with Conductor, signed policy distribution, remote kill, and free verifier access.
- [About](https://pipelab.org/about/): Company origin and the relationship between PipeLab, Pipelock, and the gauntlet.
- [Learn](https://pipelab.org/learn/): Setup guides and deep technical pages on MCP security.
- [Compare](https://pipelab.org/compare/): Comparison hub for agent security categories and competing tools.
- [Gauntlet](https://pipelab.org/gauntlet/): Public benchmark publishing Agent Egress Bench results against production binaries.

## Key reference pages

- [Agent firewall](https://pipelab.org/agent-firewall/): Canonical definition of an agent firewall vs WAFs, gateways, guardrails, and sandboxes.
- [MCP security](https://pipelab.org/learn/mcp-security/): Umbrella reference for MCP threat categories and runtime defenses.
- [MCP proxy](https://pipelab.org/learn/mcp-proxy/): Why MCP needs protocol-aware proxying and what bidirectional inspection scans.
- [Learn-and-lock](https://pipelab.org/learn/learn-and-lock/): Observed agent traffic compiled into signed contracts, shadow replay, ratification, promotion, and rollback.
- [Agent egress security](https://pipelab.org/learn/agent-egress-security/): How agents leak credentials through HTTP, DNS, and MCP.
- [MCP tool poisoning](https://pipelab.org/learn/mcp-tool-poisoning/): Malicious tool descriptions, rug-pulls, and runtime proxy-layer defense.

## High-intent guides

- [Open-source AI firewall comparison](https://pipelab.org/learn/open-source-ai-firewall/): Compares inference firewalls, agent egress firewalls, and AI gateways by enforcement boundary.
- [AI egress proxy](https://pipelab.org/learn/ai-egress-proxy/): Explains outbound agent traffic inspection, deployment patterns, and bypass prevention.
- [Verify a Pipelock receipt](https://pipelab.org/learn/verify-a-receipt/): Copy-paste verification against the public signed-receipt conformance corpus.