Skip to content

The fastest, cheapest, most efficient open-source agent harness. Run more than 500 agents on a $10 VPS.

docs(agent-harness): remove redundant runtime graph label

Steven Enamakel5b10701e59a6
diff --git a/docs/gitbooks/en/developing/architecture/agent-harness.md b/docs/gitbooks/en/developing/architecture/agent-harness.md
index 98e41cde3a..3919e80055 100644
--- a/docs/gitbooks/en/developing/architecture/agent-harness.md
+++ b/docs/gitbooks/en/developing/architecture/agent-harness.md
@@ -103,7 +103,7 @@ Multi-agent orchestration runs on the tinyagents graph layer, through `tinyagent
- The delegation graph is a `plan → execute ⇄ review → finalize` `CompiledGraph`. It has conditional routing, `RecursionPolicy`, a durable `FileCheckpointer`, a `CancellationToken` and a `GraphTracingSink`.
- The workflow phase engine fans each phase's agents out on the graph (`with_max_concurrency`). The durable `WorkflowRun` ledger stays the source of truth for resume.
- `spawn_parallel_agents` runs its fan-out through the host `spawn_parallel` adapter over `tinyagents_orchestration::parallel`.
-- The agent-teams member runtime is a conditional-routing graph (`execute → complete | fail → done`, the agent-teams runtime graph).
+- The agent-teams member runtime is a conditional-routing graph (`execute → complete | fail → done`).
- The detached-sub-agent registry is backed by a typed `TaskStore` lifecycle ledger (Pending, Running, then Completed, Failed or Cancelled).
## TinyAgents crate: features and compatibility