TAG
Trust infrastructure for AI agents — reliability, reputation, endpoint verification, and pre-invocation trust checks via MCP.
Move both the checklist and the verdict outside the model. Slot-based preconditions for irreversible tool calls — execution reads a verdict record, never a conditional.
ThumbGate Pre-Action Checks self-improve from ranked lessons and repeated failures, hard-block detected secret leaks, and block matches in strict mode.
