Preloop vs AWS Bedrock AgentCore: The open-source AI agent control plane

Preloop is the open-source AI agent control plane — an MCP firewall, AI model gateway, human approval system, and audit trail that you can self-host. AWS Bedrock AgentCore is a managed bundle of services from AWS for building and running AI agents inside the AWS platform. The honest one-liner: if you are already deep in AWS and want a managed agent runtime, AgentCore is the path of least resistance; if you want a vendor-neutral, open-source control plane that governs agents wherever they run, choose Preloop.

At a glance

What AWS Bedrock AgentCore does well

AWS Bedrock AgentCore is a credible choice when your team is already committed to AWS. It is commonly described as a bundle of seven managed services — Runtime, Memory, Gateway, Identity, Code Interpreter, Browser, and Observability — designed to cover the lifecycle of AI agents inside the AWS platform.

• Deep AWS integration. Billing, IAM, CloudWatch, VPC, and secrets management all connect with zero extra work if your workloads already run on AWS.
• Managed infrastructure. You do not operate a separate control plane. AWS runs the services.
• Framework-agnostic runtime. AgentCore positions itself as usable from LangGraph, CrewAI, LlamaIndex, Strands, and other agent frameworks.
• First-class access to Bedrock models. If you are standardizing on Claude, Nova, Titan, Llama, or Mistral through Bedrock, model routing is native.
• Enterprise trust surface. AWS security, compliance certifications, and procurement are already accepted at most enterprises.

If those attributes describe your stack, AgentCore is a reasonable starting point.

Where Preloop is different

Preloop is a different shape of product: an open-source control plane that sits in front of agents you already run, wherever they run.

• Open source and self-hostable. Preloop's core is Apache 2.0 licensed. You can run it in your own VPC, on-prem, or on any cloud. There is no forced dependency on a managed service, and there is no per-token markup on your model spend when you self-host.
• MCP-native from the ground up. Preloop treats the Model Context Protocol as its primary interface. The MCP Firewall inspects every tool call, evaluates it against policy, and decides whether to allow, deny, require approval, or require justification.
• preloop agents discover for local agents. One command scans a developer's machine, finds local Claude Code, Codex CLI, Cursor, Gemini CLI, OpenClaw, and OpenCode configurations, backs them up, and rewrites them to route through Preloop. No SDK changes. No per-repo setup.
• Human approvals that do not kill velocity. When a rule requires review, the right people get paged on mobile, Apple Watch, Slack, Mattermost, email, or a custom webhook — with full tool arguments and agent reasoning. Low-risk actions still run at full speed. Long-running reviews can use async-safe mode instead of blocking the agent's transport.
• One runtime view across every agent. Runtime session observability, model spend attribution, and tool-call audit trails live in the same product. You do not stitch an AI gateway, an MCP proxy, an approvals tool, and an observability vendor together.
• Policy-as-code in YAML + CEL. Rules are ordered, version-controlled, and live next to the rest of your infrastructure code. Conditions can inspect tool arguments, not just tool names.

When to choose AWS Bedrock AgentCore

• You are all-in on AWS Bedrock models and want to stay inside a single billing and IAM boundary.
• You prefer a fully managed service and do not want to operate any part of the control plane yourself.
• Your primary agent workloads are server-side, built on LangGraph/CrewAI/Strands, and already deployed to AWS.

When to choose Preloop

• Platform and DevEx teams rolling out Claude Code, Codex CLI, Cursor, Gemini CLI, OpenClaw, or OpenCode to engineers, who need to govern what those local agents can do without changing developer workflows.
• Security teams in regulated SaaS that need an on-prem or VPC-deployable control plane with policy-as-code, approval workflows, and searchable audit evidence.
• Engineering teams with existing multi-cloud or on-prem infrastructure who do not want to push every agent action through AWS.
• Teams building EU AI Act readiness evidence that want the operational artifacts — matched policy, approver, inputs, outcomes — in a self-hostable platform.
• Open-source-first organizations that need a control plane they can read, fork, and audit rather than a black-box managed service.
• Teams looking for an open-source alternative to AWS Bedrock AgentCore without losing approvals, gateway, or observability.

Adopting Preloop from an AgentCore world

Preloop can coexist with AWS Bedrock AgentCore. If your agents use Bedrock models, Preloop's model gateway is compatible with OpenAI and Anthropic-style APIs and can be configured to route through Bedrock endpoints at the provider layer — so model traffic keeps flowing to AWS while approvals, tool-call policy, and audit trails are enforced by Preloop.

On the local side, preloop agents discover enrolls Claude Code, Codex CLI, Cursor, Gemini CLI, OpenClaw, and OpenCode sessions on developer machines in seconds. A common migration path is: keep server-side agents on AgentCore, put developer-side coding agents under Preloop, and unify the audit and approval story through Preloop's runtime sessions view.

FAQ

Is there an open-source alternative to AWS Bedrock AgentCore? Yes. Preloop is an open-source AI agent control plane released under Apache 2.0. It covers the same core jobs — runtime visibility, gateway, policy, identity-style access control, and observability — but is self-hostable and MCP-native rather than tied to AWS.

Can I run AgentCore outside AWS? No. AWS Bedrock AgentCore is a managed AWS service. If you need a control plane that runs in your own VPC, on-prem, or on another cloud, an open-source option such as Preloop is a better fit.

Does Preloop work with Bedrock models? Yes. Preloop's AI model gateway is OpenAI/Anthropic-compatible and can be wired to Amazon Bedrock endpoints, so your teams can keep using Claude or other Bedrock-hosted models while Preloop handles tool governance, approvals, attribution, and audit.

How do I migrate from AgentCore to Preloop? For server-side agents, point your agent framework at Preloop's model gateway URL and register your MCP servers in Preloop. For developer-side agents like Claude Code, Codex CLI, or Cursor, run preloop agents discover to enroll them automatically. You can migrate incrementally — most teams start with observability and approvals, then move policy enforcement.

Does Preloop lock me into a vendor? No. The core is Apache 2.0 and self-hostable. You can run Preloop on your own infrastructure, read and audit the code, and swap model providers or MCP backends at any time.

Which AI coding agents does Preloop govern? Claude Code, Codex CLI, Cursor, Windsurf, Gemini CLI, Cline, OpenClaw, OpenCode, and any other MCP-compatible client or runtime.

Try Preloop

• Start your free trial
• Book a demo
• Preloop on GitHub