getaegis/aegis

Overview

A local-first transparent proxy that isolates credentials from AI agents. It intercepts API calls so the agent never directly sees your API keys, with no code changes required.

Best for

Best for
Developers who want to safely run third-party or untrusted AI agents on their own machine

Use cases

• Run AI agents without exposing API keys in prompts or logs
• Add credential isolation to existing agent workflows
• Enforce key rotation or access policies per agent session

Notes

A local-first transparent proxy that isolates credentials from AI agents. It intercepts API calls so the agent never directly sees your API keys, with no code changes required.

11 stars on GitHub. Last updated 2026-04-15. Licensed Apache-2.0.

Use cases

• Run AI agents without exposing API keys in prompts or logs
• Add credential isolation to existing agent workflows
• Enforce key rotation or access policies per agent session

Pros

• Simple setup with zero code modification
• Keeps credentials out of agent context and history
• Open source with a small TypeScript codebase

Cons

• Limited to agents using HTTP-based API calls
• Still early stage with only 11 GitHub stars
• Requires running a local proxy server alongside your agent

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.