5/25/2026
•
EN
Interlocks: proving the agent read the skill
How incident.io uses interlocks to prove Claude read skill files before calling tools, improving AI agent reliability.