What should your AI agent be allowed to do next?

We are sharing AEGIS on behalf of Escape Studio.

AEGIS helps protect local AI-agent workflows through security review, selected-process controls and policy checks on explicitly configured command-line and MCP routes.

- Review a selected local folder or agent profile before use. Static checks surface suspicious patterns in skills and project settings; saved snapshots help compare changes.
- Block policy-denied commands before they start. Explicitly configured AEGIS execution and selected-action routes enforce allow/deny policies. Separately, the validating MCP gateway requires exact one-use approval for a tool and its arguments.
- Respond to the process you are investigating. Suspend or resume a currently observed selected process, or stop it with explicit confirmation. AEGIS rechecks its identity before acting.
- Investigate sensitive activity. Follow a detected agent's process tree, inspect configured file observations and TCP endpoints, and open an alert with its captured evidence.

CPU/RAM charts, Claude Code token subtotals and estimated costs, audit history and exports help keep a session understandable.

The MIT-licensed Windows beta is available now. The gallery uses an editorial illustration and source UI previews with simulated data captured on 11 October 2026.

Useful feedback: which skill findings are useful, which action needs a policy control, and what evidence makes an alert easier to investigate. Source, installer verification and contribution guides are linked from the product page.

Launch copy prepared with AI assistance and the maintainer's authorization.

AEGIS

Protect local AI workflows with review and action controls

Protect local AI workflows with review and action controls