Aegis FAQs

Question 1

What is Aegis MCP?

Accepted Answer

Aegis MCP is a governance proxy that sits between AI agents and tool servers, enforcing programmatic hard constraints like rate limits, access control, and human approvals on agent actions without requiring any code changes to the agents themselves.

Question 2

Is Aegis compatible with any AI agent?

Accepted Answer

Aegis works with any MCP-compatible agent (e.g., Claude Code, OpenClaw, custom agents) because it operates at the protocol level as a transparent proxy rather than requiring SDK or framework integration.

Question 3

Does Aegis provide an audit trail for AI agent actions?

Accepted Answer

Yes, Aegis records every tool call in a persistent SQLite audit log. This includes the agent, tool, arguments, ACL/rate-limit/approval verdicts, queue position, execution duration, and the final result for complete traceability.

Question 4

How does Aegis ensure AI agents follow rules?

Accepted Answer

Aegis operates as a transparent, protocol-level proxy. Instead of relying on 'soft rules' (prompts), Aegis converts them into 'hard constraints' that agents simply cannot bypass, regardless of their internal logic or decisions.

Question 5

What key security and control features does Aegis offer?

Accepted Answer

Key features include Access Control (ACL) with per-agent/tool rules, Two-Level Rate Limiting (individual & global), Human Approval Workflows for sensitive actions, a FIFO Execution Queue with delays, and comprehensive Audit Logging of all agent activities.

Aegis

Aegis

주요 기능

사용 사례

주요 기능

사용 사례