Add roster-of-four-agents tutorial (#25)

Runs four specialized agents in a single Inngest-orchestrated workflow
with typed handoffs at every boundary, so a malformed step never
contaminates the next agent's input. Validates against the eight-check
rubric (8/8). Aligns with TBW Vol 14 "Build The Agent Org".

Co-authored-by: tbw-ai <support@getagent100.ai>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: 3 people

entries/roster-of-four-agents/README.md

@@ -0,0 +1,23 @@
+# Run a roster of four agents in a single workflow without the wheels coming off
+
+An agent team with well-defined handoffs produces the same multi-step result end-to-end in one minute that a four-person human team takes a full meeting to coordinate, and the agent team doesn't need a meeting.
+
+This is part of [AI Building Tutorials](https://github.com/thebuilderweekly/ai-building-tutorials) by [The Builder Weekly](https://thebuilderweekly.com).
+
+**Read this tutorial:**
+- [In this repo](./tutorial.md) — the raw markdown with code blocks
+- [On the web](https://thebuilderweekly.com/tutorials/roster-of-four-agents) — rendered with diagrams and syntax highlighting
+
+## What this tutorial teaches
+
+**Before:** You tried to chain four agents together and the output of each one contaminated the input of the next until the final agent produced garbage.
+
+**After:** Four agents run in a defined sequence with typed handoffs, each stage produces verifiable intermediate output, and you can inspect any stage independently if the final result looks wrong.
+
+## Tools used
+
+anthropic-api, inngest
+
+## Pillar
+
+[Agent Teams](https://thebuilderweekly.com/tutorials/pillars/agent-teams)

entries/roster-of-four-agents/metadata.json

@@ -0,0 +1,24 @@
+{
+  "id": "roster-of-four-agents",
+  "title": "Run a roster of four agents in a single workflow without the wheels coming off",
+  "slug": "roster-of-four-agents",
+  "pillar": "agent-teams",
+  "clusterTags": [
+    "multi-agent",
+    "orchestration",
+    "workflow"
+  ],
+  "soulLine": "An agent team with well-defined handoffs produces the same multi-step result end-to-end in one minute that a four-person human team takes a full meeting to coordinate, and the agent team doesn't need a meeting.",
+  "beforeState": "You tried to chain four agents together and the output of each one contaminated the input of the next until the final agent produced garbage.",
+  "afterState": "Four agents run in a defined sequence with typed handoffs, each stage produces verifiable intermediate output, and you can inspect any stage independently if the final result looks wrong.",
+  "status": "drafted",
+  "author": "tbw-ai",
+  "contributors": [],
+  "tools": [
+    "anthropic-api",
+    "inngest"
+  ],
+  "createdAt": "2026-04-29",
+  "lastVerifiedAt": "2026-04-29",
+  "freshnessWindowDays": 90
+}