@xynogen/pix-subagent 0.1.0 → 0.1.2

package/README.md

@@ -4,14 +4,16 @@ Pi extension — planner-driven sub-agents with 3 tools, live widget (model alwa
 
 ## Install
 
-Bundled by [`@xynogen/pix-core`](https://www.npmjs.com/package/@xynogen/pix-core) — no separate install needed.
-
-Standalone:
-
 ```bash
 pi install npm:@xynogen/pix-subagent
 ```
 
+> Also included in [`@xynogen/pix-core`](https://www.npmjs.com/package/@xynogen/pix-core):
+>
+> ```bash
+> pi install npm:@xynogen/pix-core
+> ```
+
 ## What it does
 
 Gives the parent agent (planner) three tools to delegate work to isolated child sessions:

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@xynogen/pix-subagent",
-	"version": "0.1.0",
+	"version": "0.1.2",
 	"description": "Pi tool — planner-driven sub-agents: spawn, fetch, steer scoped child agents",
 	"type": "module",
 	"main": "src/index.ts",

package/src/index.ts

@@ -191,9 +191,9 @@ export default function registerPixSubagent(pi: ExtensionAPI): void {
 		}
 	});
 
-	pi.on("tool_execution_start", (_event, ctx) => {
-		widget.setUICtx(ctx.ui as Parameters<typeof widget.setUICtx>[0]);
+	pi.on("turn_start", (_event, ctx) => {
 		widget.onTurnStart();
+		widget.setUICtx(ctx.ui as Parameters<typeof widget.setUICtx>[0]);
 		widget.ensureTimer();
 	});
 

package/src/tools.ts

@@ -180,10 +180,10 @@ If the target is already known, use a direct tool — \`read\` for a known path,
 
 ## Usage notes
 - Always include a short (3-5 word) description summarizing what the agent will do (shown in UI).
-- When you launch multiple agents for independent work, send them in a single message with multiple tool uses, with run_in_background: true on each, so they run concurrently.
+- When you launch multiple agents for independent work, send them in a single message with multiple tool uses, so they run concurrently.
 - When the agent is done, it returns a single message. The result is not visible to the user — to show the user, send a text message with a concise summary.
 - Trust but verify: an agent's summary describes what it intended to do, not what it did. When an agent writes or edits code, check the actual changes before reporting work as done.
-- Use run_in_background for work you don't need immediately. You will be notified when it completes — do NOT poll or sleep waiting for it.
+- Use run_in_background: true only when you explicitly do not need to see the output inline (e.g. fire-and-forget background tasks). Default is foreground — the agent streams inline and you see its work as it runs.
 - Use resume with an agent ID to continue a previous agent's work.
 - Use agent_steer to send mid-run messages to a running background agent.
 - Use model to specify a model from the available models list above (provider/id or fuzzy e.g. "haiku").