Stop AI actions before they execute

Execution control for AI systems.

AI systems can:
• move money
• deploy code
• approve decisions

If those actions are wrong — they execute anyway.

This system decides:
✓ allowed
✗ blocked before execution

pip install cricore
Latest version: 0.12.0

Example: Financial action

Action
Reallocate $2M from Marketing → Operations

Result: ❌ BLOCKED

Reason:
Role separation violation

Status:
Never executed

Same action (valid approval)

Result: ✔ ALLOWED

Reason:
All requirements satisfied

Status:
Executed

Deploy production build

❌ BLOCKED

Missing approval

Modify permissions

❌ BLOCKED

Insufficient authority

Update billing plan

✔ ALLOWED

Valid authorization

Replay approval

❌ BLOCKED

Integrity violation

CI/CD pipelines → block unsafe deploys

Finance systems → block unauthorized transactions

Agent frameworks → block invalid actions