title: "Tuskira launches open-source AI Agent Gateway" slug: "tuskira-launches-open-source-ai-agent-gateway" published: "2026-10-18" beat: "Tools" tags: ["Tools", "Launches"] creator: "Agentry Newsroom" editor: "Susanne Sperling, Editor — Human in the Loop" tools: ["Claude (Anthropic)", "Perplexity Sonar"] creativeWorkStatus: "verified" dateReviewed: "2026-10-18" aiActArticle50: "compliant" humanView: "https://agentry.news/launches/tuskira-launches-open-source-ai-agent-gateway" agentView: "https://agentry.news/agent/tuskira-launches-open-source-ai-agent-gateway"
Tuskira released the AI Agent Gateway as an open-source, self-hosted project on October 1, 2026, giving developers and enterprise teams control over agent authentication, tool access, policy enforceme
Drafted by an AI agent. Verified by Susanne Sperling, Editor — Human in the Loop. AI policy.
Tuskira released the AI Agent Gateway as an open-source project on October 1, 2026, in San Francisco, according to Business Wire. The gateway is distributed under the Apache License 2.0 and runs inside an organization's own infrastructure without requiring a Tuskira account, making it a self-hosted alternative to cloud-based agent management platforms.
The gateway sits between AI agents, MCP (Model Context Protocol) tool servers, and large-language-model providers, intercepting and controlling the flow of requests and credentials. According to Tuskira's documentation, the system authenticates requests, injects per-request credentials, scopes tools by agent profile, and logs all calls for compliance and debugging.
The project supports multiple LLM providers—Claude, AWS Bedrock, OpenAI, Gemini, and MCP tool servers—allowing organizations to switch models and tools without modifying agent code. Version 0.4.0 ships with policy controls, access profiles, comprehensive logging, token tracking, and model routing capabilities.
The gateway addresses three overlapping needs in agent deployment: developer velocity, operational control, and security governance. Developers can rapidly prototype and iterate without managing credentials or access policies manually. Platform teams gain centralized visibility into agent behavior, token spend, and tool usage across their organization. Security and compliance teams can enforce role-based access controls, audit trails, and tool restrictions at the gateway layer rather than rebuilding enforcement into each agent.
The project is labeled pre-1.0 alpha, meaning APIs and configuration may change between minor versions. Tuskira's GitHub repository includes CI/CD workflows, security documentation, and architecture specifications for self-hosted deployment.
This release reflects a wider shift in the agent economy toward infrastructure-level tooling. As agents move from experimental chatbots to production systems controlling APIs, databases, and external services, organizations need runtime gateways to enforce policy, track spend, and maintain security posture. Tuskira's open-source approach lets teams avoid vendor lock-in while adopting standardized agent governance patterns.
The gateway is available for immediate download and self-hosting on any infrastructure supporting Docker containers.