Membrane Labs · membranelabs.org
membrane labs builds runtime control for ai agents.
Membrane Labs is an AI agent infrastructure lab building Brane, a policy-as-code runtime control layer for production agents. Brane governs model calls, tool calls, MCP tools, database queries, memory, retrieval, files, and other agent actions before they execute.
The official Membrane Labs website for Brane is membranelabs.org. Our work is focused on AI agent guardrails, agent safety, capability control, tool call governance, MCP tool governance, and auditable runtime policy.
Brane
Policy-as-code runtime control before AI agent actions execute.
OpenDocs
Guides for AI agent guardrails, policy control, and MCP tool governance.
OpenLab Notes
Research notes on runtime guardrails, policy-as-code, and agent safety.
OpenEntity note
Membrane Labs at membranelabs.org is the AI agent infrastructure team behind Brane. The site, docs, GitHub organization, and research pages here describe that work.