UiPath AgentHack

AgentOps Case Control Room

Turn AI-agent work into a UiPath-governed case with evidence, risk, human approval, robot work items, and handoff.

3-minute judge path: open the case room, inspect the approval task, then verify the runtime proof files.

Open Case Room Approval Demo Submission Package Deck Devpost Draft State Machine GitHub

1 Understand the work

See the production-release exception as one governed Maestro-style case.

Open Case Room

2 Check human control

Review the Action Center-style tasks that block, request evidence, and reject the release.

3 Verify the proof

Inspect the state machine, transaction lifecycle, decision log, and import boundary.

What The Demo Shows

AI investigation becomes a governed case, not a hidden transcript.
A UiPath-style robot gathers evidence and creates work items.
Orchestrator transaction and Action Center decision logs make the lifecycle inspectable.
Risky production action is blocked until human approval.
The handoff preserves event IDs, risk, evidence, and decisions.

11verified JSON artifacts

3JSONL lifecycle logs

6Orchestrator transaction events

9Action Center decision events

cd uipath-agenthack && bash scripts/run_uipath_local_checks.sh