agent-swarm.devHot Patterns
Five patterns that recur across every playbook — litmus tests, drain loops, HITL gates, per-customer working directories, and no-op workflows. These are the recipes that compound, regardless of which use case you're building.
These are the patterns that show up in nearly every playbook. Treat them as recipes — short, focused, opinionated — that you can pull into any new flow you build on top of the swarm.
- Litmus Tests — LLM-as-judge quality gates that hard-reject sub-bar output, using a different model family from the generator.
- Drain Loops — turn one big ticket into a chain of stacked, individually-reviewable PRs, then a merge loop that halts on the first failure.
- HITL Gates — pause a workflow mid-flight until a human approves in Slack. The agent does the work; the human owns the irreversible step.
- Per-Customer Working Directories — give each top account a persistent folder in agent-fs so context compounds across months.
- No-op When Nothing Changed — detect "did anything actually happen?" and skip silently when not. Default for any scheduled job.
Open-Prompt Showreel Videos
Make design-led motion-graphics videos with an agent swarm from one open prompt. One Claude Code worker on Opus at max effort designs, codes, renders and critiques the video; every number on screen is read live from a named source. Includes the full toolchain setup and a worked example, the TLA+ races showreel.
Pattern: Litmus Tests (LLM-as-Judge Quality Gates)
A quality gate where one agent hard-rejects another agent's output against an explicit rubric — using a different model family so the judgment is genuinely independent and not rubber-stamping in disguise.