Free AI Agent Risk Check

Find out what your AI agent can already reach.

Answer eight questions or connect a repo reference. You get a partial score immediately; the licensed scan unlocks findings, preventive controls, policy files, and the PDF report.

Work emailCompany

Repository URL or upload referencePartner or campaign code

Use a repo URL or private upload reference. The free check stores the scope only; licensed scans require proof of ownership or written authorization.

Do AI coding tools have access to your full repository?

Can agents run terminal commands in the project?

Are secrets or production configs stored near application code?

Can agents edit GitHub workflows, CI files, or deployment scripts?

Do docs, issues, or prompts contain operational instructions?

Do you review agent-created changes before merge?

Do you have a policy file for Codex, Claude, Cursor, Copilot, or internal agents?

Do you monitor new files that expand AI-agent access?

Licensed scan intake

Submit authorized scope after purchase.

CapitalGuard turns the license into a controlled scan handoff: payment reference, repo authorization, AI tool permissions, report recipient, and monitoring preference in one packet. The service detects exposure and installs prevention guidance; it does not claim to repair the codebase for you.