AgentPulse, drift investigation for multi vs Clay Seal

Open-source drift investigation for multi-agent systems: which agent drifted, where it originated, and what to check next.

Runtime security for AI agents. Clay Seal narrows what an agent can do as the task changes. Open-source identity and receipts; design-partner runtime sandbox.

About AgentPulse, drift investigation for multi

Open-source drift investigation for multi-agent systems: which agent drifted, where it originated, and what to check next. What did change is the payload it receives from writer, starting at the same run. So critic is probably downstream damage, not the source. After the change, writer responses run ~180% longer and drop the section structure critic expects, earlier payloads were structured summaries, newer ones read as prose. If restoring the format fixes critic, make the fix in writer.

About Clay Seal

Runtime security for AI agents. Clay Seal narrows what an agent can do as the task changes. Open-source identity and receipts; design-partner runtime sandbox. Each action is checked against what the task needs now, not what the agent received at launch. The permissions an agent starts with are a guess about everything it might ever need. The walls move in as the task takes shape, and anything that reaches outside them is blocked, no matter what the original grant allowed.