Tuskira launches AI Agent Gateway to keep credentials out of agent configs

Tuskira has released AI Agent Gateway, an open-source tool that sits between AI agents and the MCP servers and model providers they call. The gateway verifies each agent's key and profile, injects real credentials from an encrypted store, and blocks unauthorized tool calls. It runs on macOS and Linux and is free on GitHub.

• Gateway checks the agent's key and profile on every tool call
• Real tokens stay in an encrypted store and never reach agent configs
• Supports Anthropic, AWS Bedrock, OpenAI and Gemini with token and cost tracking
• Denied calls return an error and are logged without reaching the backend