Random Agent Chain
Demonstrates dynamic routing — agents chosen on the fly, gated in the queue.
What it does
A single observation raises one approval. Each approval has the system pick an agent at random from those that haven't run, enrich the shared context, and — if any agent remains — raise the next approval. Approve to continue; reject to stop.
How it works
A single observation whose approval chains itself: on each approval the system picks one agent at random from those that haven't contributed, that agent enriches the shared context, and — if any agent remains — it raises the next approval. Approve to keep going; reject to stop. Shows LLM-routed, human-gated context passing across three agents from ONE observation.
The team
Import into StudioX
- Download the
.sxmfile. - In StudioX → Missions → Import, drop the file.
- Review, finalize, and it's live — chat to it or enable the observation.
Reviews
More in Patterns · Demos
Bug Analysis Chain
Demonstrates a three-stage handoff — analyze, pull the logs, propose a fix.
Bug Analysis Agent · Log Analysis Agent · Code Fixing Agent
Context Passing Chain
Demonstrates context flowing cleanly from one agent to the next.
Agent A · Agent B · Agent C