TOOL
OpenShell runs each AI agent in an isolated sandbox where a policy decides every file, system call and network connection it can make. Agents never see real credentials; OpenShell adds them only to requests bound for approved endpoints, and risky policy changes wait for human review. That is the kind of hard fence agents need before they go near plant networks.
For IT and OT security leads: piloting agents inside plant or utility networks.
Found via @RoundtableSpace on X
More from The Feed
TOOL
An MIT-licensed Python reference turns telemetry, alarms, an asset ontology and maintenance knowledge into answers tied to a specific asset and alarm, with guardrails that block ungrounded numbers and unsafe writes. It scores those answers with a repeatable eval harness. It says plainly that operators want which asset and which alarm, not a fluent paragraph.
For reliability and controls engineers: testing an AI assistant on plant data.
TOOL
plc-manual-mcp turns a bookmarked PLC manual PDF, with the Siemens S7-1200 as the worked example, into a full-text database served over MCP. A coding assistant can then look up exact instruction syntax, parameter tables and data types while writing PLC code. No manual content ships; you build it from your own licensed copy.
For controls engineers and PLC programmers: using Claude Code or Cursor.
TOOL
An agent skill walks an AI assistant from stakeholder values through soft systems methods, STPA hazard analysis and requirements quality scoring, with human gates and an audit trail. Hazard analysis is where an agent that shows its work and stops for a human sign-off makes sense. Outputs still need an engineer's review.
For process safety and systems engineers: writing requirements or running STPA.