diff --git a/prompts/tools-cascade.md b/prompts/tools-cascade.md
index 7490220..555d1b3 100644
--- a/prompts/tools-cascade.md
+++ b/prompts/tools-cascade.md
@@ -11,13 +11,11 @@ Delegate mechanical subtasks to a cheaper/weaker model. Fire-and-forget: returns
 cascade: instruction="<fully self-contained task description>" [id="<optional-name>"]
 ```
 
-Parallel — use `{parallel: [...]}`:
+Parallel — issue multiple cascade calls in the same step; they execute concurrently:
 ```
-execute_code: steps=[{parallel: [
-  {code: "cascade 'Read /src/a.go and list all exported functions'"},
-  {code: "cascade 'Read /src/b.go and list all exported functions'"},
-  {code: "cascade 'Read /src/c.go and list all exported functions'"}
-]}]
+cascade 'Read /src/a.go and list all exported functions'
+cascade 'Read /src/b.go and list all exported functions'
+cascade 'Read /src/c.go and list all exported functions'
 ```
 
 **Returns** (KV, one per line):
@@ -41,6 +39,7 @@ You will receive this as your next queued turn. Correlate via the `[from=...]` h
 
 - **Instructions must be fully self-contained.** The worker has zero context beyond what you put in the instruction string. Include file paths, expected formats, relevant snippets — everything it needs.
 - **Never wait for results.** After calling cascade, continue your own work or end your turn. You will be woken when results arrive.
+- **Results arrive asynchronously.** Because the callback is a queued turn, cascade is only appropriate for work whose result you do not need in the *current* turn. If you need the answer right now to proceed, do the work yourself.
 - **One task per call.** Do not bundle multiple unrelated tasks into one instruction.
 - **Parallelism is safe.** cascade is marked `ollie:parallel read` — multiple calls execute concurrently.