switchroom 0.15.2 → 0.15.4

package/telegram-plugin/uat/scenarios/jtbd-model-command-dm.test.ts

@@ -0,0 +1,93 @@
+/**
+ * UAT — `/model` Telegram command (PR #2259, shipped v0.15.2).
+ *
+ * Serves: `reference/vision.md` outcome 2 (you hold the leash) — the
+ * operator can see and switch the agent's Claude model from Telegram
+ * without SSH. Session-scoped switch via claude's own `/model <name>`
+ * REPL verb injected into the tmux pane.
+ *
+ * Three assertions against a real agent over real Telegram:
+ *
+ *   1. Bare `/model` → shows the configured model (never opens claude's
+ *      interactive picker — the reply must come from the gateway, fast,
+ *      containing "Configured:").
+ *   2. `/model <valid-name>` → switch is injected; reply relays claude's
+ *      response and carries the session-only persistence note.
+ *   3. `/model bogus-name` → still a reply (claude's inline error is
+ *      relayed, or the empty-capture explanation) — never silence.
+ *
+ * The switch test sets the model to the SAME value the agent already
+ * runs (sonnet) so the canary doesn't leave the harness agent on a
+ * different model afterwards.
+ */
+
+import { describe, it, expect } from "vitest";
+import { spinUp } from "../harness.js";
+
+const AGENT = "test-harness";
+const REPLY_TIMEOUT_MS = 30_000;
+
+describe("uat: /model command — show, switch, bad-name", () => {
+  it(
+    "bare /model shows the model dashboard (menu v2) or static fallback (v1)",
+    async () => {
+      const sc = await spinUp({ agent: AGENT });
+      try {
+        await sc.sendDM("/model");
+        // v2 (picker-driven menu): "Now: <model>"; v1 / fallback path:
+        // "Configured: <model>". Either proves the gateway handled the
+        // command rather than forwarding it to claude as plain text.
+        const shape = /Now:|Configured:/i;
+        const reply = await sc.expectMessage(shape, {
+          from: "bot",
+          timeout: REPLY_TIMEOUT_MS,
+        });
+        expect(reply.text).toMatch(shape);
+        // Persistence caveat present on both shapes
+        expect(reply.text).toMatch(/switchroom\.yaml/i);
+      } finally {
+        await sc.tearDown();
+      }
+    },
+    60_000,
+  );
+
+  it(
+    "/model sonnet switches the live session (same-value, no net change)",
+    async () => {
+      const sc = await spinUp({ agent: AGENT });
+      try {
+        await sc.sendDM("/model sonnet");
+        // Accept either a relayed claude response or the explicit
+        // empty-capture explanation — both prove the command routed
+        // through the gateway handler (and neither is silence).
+        const reply = await sc.expectMessage(
+          /\/model sonnet|no response captured|Session-only/i,
+          { from: "bot", timeout: REPLY_TIMEOUT_MS },
+        );
+        expect(reply.text.length).toBeGreaterThan(0);
+      } finally {
+        await sc.tearDown();
+      }
+    },
+    60_000,
+  );
+
+  it(
+    "/model bogus-name still gets a reply (error relayed, never silence)",
+    async () => {
+      const sc = await spinUp({ agent: AGENT });
+      try {
+        await sc.sendDM("/model bogus-model-name-xyz");
+        const reply = await sc.expectMessage(/\S/, {
+          from: "bot",
+          timeout: REPLY_TIMEOUT_MS,
+        });
+        expect(reply.text.length).toBeGreaterThan(0);
+      } finally {
+        await sc.tearDown();
+      }
+    },
+    60_000,
+  );
+});

package/telegram-plugin/welcome-text.ts

@@ -66,6 +66,14 @@ export type StatusProbeRow = {
 export type AgentMetadata = {
   agentName: string;
   model: string | null;
+  /**
+   * Live session-model override set via the `/model` picker (session-only,
+   * resets on restart). When present it's what the agent is ACTUALLY running
+   * right now, distinct from `model` (the persistent configured model). Null
+   * when no session switch is active — then `/status` just shows `model`.
+   * Surfaced so `/status` and `/model` never silently disagree.
+   */
+  sessionModel?: string | null;
   extendsProfile: string | null;
   topicName: string | null;
   topicEmoji: string | null;
@@ -122,7 +130,14 @@ export function formatAgentLine(meta: AgentMetadata): string {
   const topic = meta.topicName
     ? ` · topic: ${escapeHtml([meta.topicEmoji, meta.topicName].filter(Boolean).join(" "))}`
     : "";
-  return `<b>${escapeHtml(meta.agentName)}</b> · model: <code>${escapeHtml(m)}</code>${topic}`;
+  // A live `/model` session switch overrides what's running. Show it next to
+  // the configured model so the two surfaces agree (the override resets on
+  // restart, when the session reverts to the configured model).
+  const session =
+    meta.sessionModel && meta.sessionModel.length > 0
+      ? ` · live session: <code>${escapeHtml(meta.sessionModel)}</code>`
+      : "";
+  return `<b>${escapeHtml(meta.agentName)}</b> · model: <code>${escapeHtml(m)}</code>${session}${topic}`;
 }
 
 /**

package/profiles/default/workspace/HEARTBEAT.md.hbs

@@ -1,40 +0,0 @@
-# HEARTBEAT.md — Proactive Check-Ins
-
-This file is read on every turn (it's a dynamic workspace file). Edit it
-to tell yourself what to look for when someone (or something) prompts you
-with a bare "heartbeat" — a cron firing, a quiet-period nudge, or a
-scheduled check-in.
-
-## When this fires
-
-A heartbeat arrives as a user-role message with no real payload — often
-just "HEARTBEAT" or "heartbeat check". When that happens:
-
-1. Run through the bullets below in order.
-2. If anything needs action, respond normally (and take the action).
-3. If nothing needs action, respond with exactly `HEARTBEAT_OK` on its
-   own line. The plugin suppresses that as a silent reply — no Telegram
-   message gets sent, and the user isn't notified.
-
-## Things to check (customize per-agent)
-
-- **New emails / messages:** is there anything in the inbox or
-  connected channels that looks actionable?
-- **Upcoming calendar events:** anything in the next ~2 hours the user
-  should be reminded of?
-- **Long-running tasks:** any background work you kicked off earlier
-  that might have completed?
-- **Today's plan:** anything in `memory/YYYY-MM-DD.md` (today's
-  daily note, auto-loaded into context by the dynamic workspace hook)
-  that hasn't been touched?
-
-## Guidelines
-
-- **Respect quiet hours.** If it's late (local time 22:00–08:00),
-  default to `HEARTBEAT_OK` unless something is genuinely urgent.
-- **Don't spam.** If you messaged the user in the last 30 minutes,
-  `HEARTBEAT_OK` unless there's something new to add.
-- **Stay terse.** A heartbeat-initiated message should be one or two
-  lines, not a paragraph.
-
-Edit this file to narrow or broaden the check set for this agent.