@cohortapp/agent-sdk 2.3.0

package/scripts/huddle/package-lock.json

@@ -0,0 +1,62 @@
+{
+  "name": "maestro-huddle",
+  "version": "1.0.0",
+  "lockfileVersion": 3,
+  "requires": true,
+  "packages": {
+    "": {
+      "name": "maestro-huddle",
+      "version": "1.0.0",
+      "dependencies": {
+        "@anthropic-ai/sdk": "^0.52.0",
+        "dotenv": "^16.5.0",
+        "ws": "^8.18.0"
+      },
+      "engines": {
+        "node": ">=20.0.0"
+      }
+    },
+    "node_modules/@anthropic-ai/sdk": {
+      "version": "0.52.0",
+      "resolved": "https://registry.npmjs.org/@anthropic-ai/sdk/-/sdk-0.52.0.tgz",
+      "integrity": "sha512-d4c+fg+xy9e46c8+YnrrgIQR45CZlAi7PwdzIfDXDM6ACxEZli1/fxhURsq30ZpMZy6LvSkr41jGq5aF5TD7rQ==",
+      "license": "MIT",
+      "bin": {
+        "anthropic-ai-sdk": "bin/cli"
+      }
+    },
+    "node_modules/dotenv": {
+      "version": "16.6.1",
+      "resolved": "https://registry.npmjs.org/dotenv/-/dotenv-16.6.1.tgz",
+      "integrity": "sha512-uBq4egWHTcTt33a72vpSG0z3HnPuIl6NqYcTrKEg2azoEyl2hpW0zqlxysq2pK9HlDIHyHyakeYaYnSAwd8bow==",
+      "license": "BSD-2-Clause",
+      "engines": {
+        "node": ">=12"
+      },
+      "funding": {
+        "url": "https://dotenvx.com"
+      }
+    },
+    "node_modules/ws": {
+      "version": "8.20.0",
+      "resolved": "https://registry.npmjs.org/ws/-/ws-8.20.0.tgz",
+      "integrity": "sha512-sAt8BhgNbzCtgGbt2OxmpuryO63ZoDk/sqaB/znQm94T4fCEsy/yV+7CdC1kJhOU9lboAEU7R3kquuycDoibVA==",
+      "license": "MIT",
+      "engines": {
+        "node": ">=10.0.0"
+      },
+      "peerDependencies": {
+        "bufferutil": "^4.0.1",
+        "utf-8-validate": ">=5.0.2"
+      },
+      "peerDependenciesMeta": {
+        "bufferutil": {
+          "optional": true
+        },
+        "utf-8-validate": {
+          "optional": true
+        }
+      }
+    }
+  }
+}

package/scripts/huddle/package.json

@@ -0,0 +1,22 @@
+{
+  "name": "maestro-huddle",
+  "version": "1.0.0",
+  "description": "Maestro — Slack huddle voice participation (agent-neutral)",
+  "main": "huddle-server.mjs",
+  "type": "module",
+  "scripts": {
+    "start": "node huddle-server.mjs",
+    "setup": "./setup-audio.sh",
+    "launch-slack": "./launch-slack.sh",
+    "join": "node huddle-server.mjs --join",
+    "call": "node huddle-server.mjs --call"
+  },
+  "engines": {
+    "node": ">=20.0.0"
+  },
+  "dependencies": {
+    "@anthropic-ai/sdk": "^0.52.0",
+    "dotenv": "^16.5.0",
+    "ws": "^8.18.0"
+  }
+}

package/scripts/huddle/setup-audio.sh

@@ -0,0 +1,239 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# ---------------------------------------------------------------------------
+# Maestro Huddle — Audio Setup
+#
+# Installs and configures BlackHole virtual audio devices for Slack huddle
+# participation. Two separate virtual devices prevent feedback loops:
+#
+#   BlackHole 2ch  → Slack's Speaker output (captures what others say)
+#   BlackHole 16ch → Slack's Mic input    (carries the agent's synthesized voice)
+#
+# Usage:
+#   ./setup-audio.sh           # Full install + verify
+#   ./setup-audio.sh --check   # Verify existing setup only
+# ---------------------------------------------------------------------------
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+LOG_FILE="${SCRIPT_DIR}/../../logs/huddle/setup-audio.log"
+
+mkdir -p "$(dirname "$LOG_FILE")"
+
+log() { echo "[$(date '+%H:%M:%S')] $*" | tee -a "$LOG_FILE"; }
+err() { echo "[$(date '+%H:%M:%S')] ERROR: $*" | tee -a "$LOG_FILE" >&2; }
+ok()  { echo "[$(date '+%H:%M:%S')] ✓ $*" | tee -a "$LOG_FILE"; }
+
+# ---------------------------------------------------------------------------
+# Pre-flight checks
+# ---------------------------------------------------------------------------
+
+check_homebrew() {
+  if ! command -v brew &>/dev/null; then
+    err "Homebrew not installed. Install from https://brew.sh"
+    exit 1
+  fi
+  ok "Homebrew available"
+}
+
+check_switchaudio() {
+  if ! command -v SwitchAudioSource &>/dev/null; then
+    log "Installing switchaudio-osx..."
+    brew install switchaudio-osx
+  fi
+  ok "SwitchAudioSource available"
+}
+
+check_sox() {
+  if ! command -v sox &>/dev/null; then
+    log "Installing sox (audio capture/playback)..."
+    brew install sox
+  fi
+  ok "sox available"
+}
+
+# ---------------------------------------------------------------------------
+# BlackHole installation
+# ---------------------------------------------------------------------------
+
+install_blackhole_2ch() {
+  if SwitchAudioSource -a 2>/dev/null | grep -q "BlackHole 2ch"; then
+    ok "BlackHole 2ch already installed"
+    return 0
+  fi
+
+  log "Installing BlackHole 2ch (huddle output capture)..."
+  brew install blackhole-2ch
+
+  # Verify installation
+  sleep 2
+  if SwitchAudioSource -a 2>/dev/null | grep -q "BlackHole 2ch"; then
+    ok "BlackHole 2ch installed successfully"
+  else
+    err "BlackHole 2ch installation failed — may need restart"
+    return 1
+  fi
+}
+
+install_blackhole_16ch() {
+  if SwitchAudioSource -a 2>/dev/null | grep -q "BlackHole 16ch"; then
+    ok "BlackHole 16ch already installed"
+    return 0
+  fi
+
+  log "Installing BlackHole 16ch (agent mic input)..."
+  brew install blackhole-16ch
+
+  sleep 2
+  if SwitchAudioSource -a 2>/dev/null | grep -q "BlackHole 16ch"; then
+    ok "BlackHole 16ch installed successfully"
+  else
+    err "BlackHole 16ch installation failed — may need restart"
+    return 1
+  fi
+}
+
+# ---------------------------------------------------------------------------
+# Verification
+# ---------------------------------------------------------------------------
+
+verify_audio_devices() {
+  log ""
+  log "=== Audio Device Verification ==="
+  log ""
+
+  local all_devices
+  all_devices=$(SwitchAudioSource -a 2>/dev/null)
+
+  log "All audio devices:"
+  echo "$all_devices" | while read -r line; do
+    log "  $line"
+  done
+
+  local pass=true
+
+  if echo "$all_devices" | grep -q "BlackHole 2ch"; then
+    ok "BlackHole 2ch present (Slack speaker → capture)"
+  else
+    err "BlackHole 2ch NOT found"
+    pass=false
+  fi
+
+  if echo "$all_devices" | grep -q "BlackHole 16ch"; then
+    ok "BlackHole 16ch present (TTS → Slack mic)"
+  else
+    err "BlackHole 16ch NOT found"
+    pass=false
+  fi
+
+  if command -v sox &>/dev/null; then
+    ok "sox available for audio capture/playback"
+  else
+    err "sox NOT found"
+    pass=false
+  fi
+
+  log ""
+
+  if [ "$pass" = true ]; then
+    ok "Audio setup verified — all devices present"
+    log ""
+    log "=== Slack Configuration Required ==="
+    log ""
+    log "Open Slack → Preferences → Audio & Video:"
+    log "  Speaker:    BlackHole 2ch"
+    log "  Microphone: BlackHole 16ch"
+    log ""
+    log "This routes huddle audio through the agent's capture pipeline."
+    log "System audio remains on built-in devices (no disruption)."
+    return 0
+  else
+    err "Audio setup incomplete — see errors above"
+    return 1
+  fi
+}
+
+# ---------------------------------------------------------------------------
+# Audio routing test
+# ---------------------------------------------------------------------------
+
+test_audio_capture() {
+  log ""
+  log "=== Audio Capture Test ==="
+  log ""
+  log "Recording 3 seconds from BlackHole 2ch..."
+
+  local test_file="/tmp/agent-audio-test-$(date +%s).wav"
+
+  # Record 3 seconds from BlackHole 2ch
+  if sox -t coreaudio "BlackHole 2ch" -r 16000 -c 1 -b 16 "$test_file" trim 0 3 2>/dev/null; then
+    local file_size
+    file_size=$(stat -f%z "$test_file" 2>/dev/null || echo "0")
+    ok "Captured ${file_size} bytes to ${test_file}"
+
+    if [ "$file_size" -gt 1000 ]; then
+      ok "Audio capture working (got data)"
+    else
+      log "Captured silence (expected if nothing is playing through BlackHole 2ch)"
+    fi
+
+    rm -f "$test_file"
+  else
+    err "Audio capture failed — BlackHole 2ch may not be accessible"
+    return 1
+  fi
+}
+
+test_audio_playback() {
+  log ""
+  log "=== Audio Playback Test ==="
+  log ""
+  log "Generating 1-second test tone on BlackHole 16ch..."
+
+  # Generate a 440Hz sine wave and play to BlackHole 16ch
+  if sox -n -t coreaudio "BlackHole 16ch" synth 1 sine 440 2>/dev/null; then
+    ok "Playback to BlackHole 16ch succeeded"
+    log "If Slack mic is set to BlackHole 16ch, this tone would be heard in a huddle"
+  else
+    err "Playback to BlackHole 16ch failed"
+    return 1
+  fi
+}
+
+# ---------------------------------------------------------------------------
+# Main
+# ---------------------------------------------------------------------------
+
+main() {
+  log ""
+  log "=========================================="
+  log "  Maestro Huddle — Audio Setup"
+  log "  $(date)"
+  log "=========================================="
+  log ""
+
+  if [ "${1:-}" = "--check" ]; then
+    check_switchaudio
+    verify_audio_devices
+    exit $?
+  fi
+
+  if [ "${1:-}" = "--test" ]; then
+    test_audio_capture
+    test_audio_playback
+    exit $?
+  fi
+
+  # Full installation
+  check_homebrew
+  check_switchaudio
+  check_sox
+  install_blackhole_2ch
+  install_blackhole_16ch
+  verify_audio_devices
+
+  log ""
+  log "Run './setup-audio.sh --test' after configuring Slack audio to verify the pipeline."
+}
+
+main "$@"

package/scripts/huddle/start-call.mjs

@@ -0,0 +1,318 @@
+#!/usr/bin/env node
+/**
+ * Start a Slack huddle call with someone and then launch the huddle server.
+ *
+ * This script handles the two-step huddle flow:
+ *   1. Click huddle button in main window (opens preview)
+ *   2. Click "Start huddle" in preview window (starts WebRTC)
+ *   3. Launch the huddle server to connect audio bridge
+ *
+ * Usage:
+ *   node start-call.mjs principal
+ *   node start-call.mjs alex
+ */
+
+import { WebSocket } from "ws";
+import { execFileSync, execFile, spawn } from "node:child_process";
+import { promisify } from "node:util";
+import { dirname, join } from "node:path";
+import { fileURLToPath } from "node:url";
+import dotenv from "dotenv";
+
+const __dirname = dirname(fileURLToPath(import.meta.url));
+dotenv.config({ path: join(__dirname, "../..", ".env") });
+
+const execFileAsync = promisify(execFile);
+const CDP_URL = "http://localhost:9222";
+const target = process.argv[2] || "principal";
+
+// ---------------------------------------------------------------------------
+// Helper: send CDP command
+// ---------------------------------------------------------------------------
+
+function cdpEval(wsUrl, expression, timeout = 5000) {
+  return new Promise((resolve, reject) => {
+    const ws = new WebSocket(wsUrl);
+    const timer = setTimeout(() => { ws.close(); reject(new Error("timeout")); }, timeout);
+    ws.on("error", (err) => { clearTimeout(timer); reject(err); });
+    ws.on("open", () => {
+      ws.send(JSON.stringify({
+        id: 1,
+        method: "Runtime.evaluate",
+        params: { expression, returnByValue: true, awaitPromise: true },
+      }));
+    });
+    ws.on("message", (raw) => {
+      const msg = JSON.parse(raw.toString());
+      if (msg.id === 1) {
+        clearTimeout(timer);
+        ws.close();
+        resolve(msg.result?.result?.value);
+      }
+    });
+  });
+}
+
+async function fetchTargets() {
+  const resp = await fetch(`${CDP_URL}/json`);
+  return resp.json();
+}
+
+function sleep(ms) {
+  return new Promise((r) => setTimeout(r, ms));
+}
+
+// ---------------------------------------------------------------------------
+// Main
+// ---------------------------------------------------------------------------
+
+async function main() {
+  console.log(`\n=== Starting huddle call with: ${target} ===\n`);
+
+  // Step 0: Find main Slack target
+  const targets = await fetchTargets();
+  const mainTarget = targets.find(
+    (t) => t.type === "page" && t.url?.includes("app.slack.com/client")
+  );
+  if (!mainTarget) {
+    console.error("ERROR: No Slack page target found. Is Slack running with CDP?");
+    process.exit(1);
+  }
+  const mainWs = mainTarget.webSocketDebuggerUrl;
+  console.log("[OK] Found Slack target:", mainTarget.title.substring(0, 50));
+
+  // Step 1: Initialize microphone access and set audio preferences.
+  // After a Slack restart, input devices aren't enumerable until
+  // getUserMedia() or askForMediaAccess() is called. We ensure both
+  // permission and device enumeration are ready before proceeding.
+  await cdpEval(
+    mainWs,
+    `(async () => {
+      const calls = window.desktop?.calls;
+      if (calls?.askForMediaAccess) await calls.askForMediaAccess('microphone');
+      try {
+        const s = await navigator.mediaDevices.getUserMedia({audio: true});
+        s.getTracks().forEach(t => t.stop());
+      } catch {}
+    })()`,
+    10000
+  );
+  console.log("[OK] Microphone access initialized");
+
+  const inputId = "9f391a3f2efb9f50f9dc8dea1534700bcc5d4e0478b56438930a2071401b25bb";
+  const outputId = "a930ae0eb95e3c597c4722e8bc232890be7f94ef6c9b864c4170a58ba28b5b6b";
+  await cdpEval(
+    mainWs,
+    `localStorage.setItem('preferred-media-devices', JSON.stringify({audioinput: '${inputId}', audiooutput: '${outputId}'}))`
+  );
+  console.log("[OK] Audio preferences set (BlackHole 16ch in, 2ch out)");
+
+  // Step 2: Navigate to target's DM (Cmd+K → type name → Enter)
+  console.log(`[..] Navigating to ${target}'s DM...`);
+  await execFileAsync("osascript", [
+    "-e",
+    `tell application "Slack" to activate
+     delay 0.5
+     tell application "System Events"
+       keystroke "k" using {command down}
+     end tell`,
+  ]);
+  await sleep(800);
+  await execFileAsync("osascript", [
+    "-e",
+    `tell application "System Events" to keystroke "${target}"`,
+  ]);
+  await sleep(1200);
+  await execFileAsync("osascript", [
+    "-e",
+    `tell application "System Events" to key code 36`,
+  ]);
+  await sleep(2000);
+  console.log("[OK] Navigated to DM");
+
+  // Step 3: Click huddle button via React props
+  console.log("[..] Opening huddle preview...");
+  const clickResult = await cdpEval(
+    mainWs,
+    `(() => {
+      const btn = document.querySelector('[data-qa="huddle_channel_header_button__start_button"]');
+      if (!btn) return 'no-button';
+      const pk = Object.keys(btn).find(k => k.startsWith('__reactProps'));
+      if (pk && btn[pk]?.onClick) {
+        btn[pk].onClick({
+          target: btn, currentTarget: btn,
+          nativeEvent: new MouseEvent('click', {bubbles: true}),
+          type: 'click', preventDefault: () => {}, stopPropagation: () => {}, persist: () => {}
+        });
+        return 'clicked';
+      }
+      return 'no-handler';
+    })()`
+  );
+  if (clickResult !== "clicked") {
+    console.error(`ERROR: Huddle button click failed: ${clickResult}`);
+    process.exit(1);
+  }
+  console.log("[OK] Huddle button clicked — waiting for preview...");
+
+  // Step 4: Wait for preview window to appear and get its position
+  let previewPos = null;
+  for (let attempt = 0; attempt < 10; attempt++) {
+    await sleep(1000);
+    try {
+      const { stdout } = await execFileAsync("osascript", [
+        "-e",
+        `tell application "System Events"
+          tell process "Slack"
+            repeat with i from 1 to (count of windows)
+              if name of window i contains "Huddle" then
+                set p to position of window i
+                set s to size of window i
+                return (item 1 of p as text) & "," & (item 2 of p as text) & "|" & (item 1 of s as text) & "x" & (item 2 of s as text)
+              end if
+            end repeat
+            return "waiting"
+          end tell
+        end tell`,
+      ]);
+      const result = stdout.trim();
+      if (result !== "waiting") {
+        const [posStr, sizeStr] = result.split("|");
+        const [wx, wy] = posStr.split(",").map(Number);
+        const [ww, wh] = sizeStr.split("x").map(Number);
+        previewPos = { x: wx, y: wy, w: ww, h: wh };
+        console.log(
+          `[OK] Preview window found at (${wx},${wy}), size ${ww}x${wh}`
+        );
+
+        // Move window up if Start button would be behind dock
+        if (wy + wh > 1000) {
+          await execFileAsync("osascript", [
+            "-e",
+            `tell application "System Events"
+              tell process "Slack"
+                repeat with i from 1 to (count of windows)
+                  if name of window i contains "Huddle" then
+                    set position of window i to {${wx}, 0}
+                  end if
+                end repeat
+              end tell
+            end tell`,
+          ]);
+          previewPos.y = 0;
+          console.log("[OK] Moved preview window up to avoid dock");
+        }
+        break;
+      }
+    } catch {
+      // Window not ready yet
+    }
+  }
+
+  if (!previewPos) {
+    console.error("ERROR: Preview window never appeared");
+    process.exit(1);
+  }
+
+  // Step 5: Find Start button in preview CDP target
+  await sleep(1000);
+  const updatedTargets = await fetchTargets();
+  const previewTarget = updatedTargets.find(
+    (t) => t.type === "page" && t.title?.includes("Huddle")
+  );
+
+  let btnX = 316,
+    btnY = previewPos.h - 70; // Default: center-bottom
+
+  if (previewTarget) {
+    try {
+      const btnPosStr = await cdpEval(
+        previewTarget.webSocketDebuggerUrl,
+        `(() => {
+          const btn = document.querySelector('[data-qa="huddle_invite_window_with_preview_content_join_button"]')
+            || Array.from(document.querySelectorAll('button')).find(b => b.getAttribute('aria-label') === 'Start huddle');
+          if (!btn) return null;
+          const r = btn.getBoundingClientRect();
+          return JSON.stringify({x: Math.round(r.x + r.width/2), y: Math.round(r.y + r.height/2)});
+        })()`
+      );
+      if (btnPosStr) {
+        const pos = JSON.parse(btnPosStr);
+        btnX = pos.x;
+        btnY = pos.y;
+        console.log(`[OK] Start button at viewport (${btnX}, ${btnY})`);
+      }
+    } catch (err) {
+      console.log(`[..] CDP button lookup failed (${err.message}), using default position`);
+    }
+  }
+
+  // Step 6: Click Start huddle via cliclick
+  const screenX = previewPos.x + btnX;
+  const screenY = previewPos.y + btnY;
+  console.log(`[..] Clicking Start huddle at screen (${screenX}, ${screenY})...`);
+
+  // Raise the preview window first
+  await execFileAsync("osascript", [
+    "-e",
+    `tell application "System Events"
+      tell process "Slack"
+        repeat with i from 1 to (count of windows)
+          if name of window i contains "Huddle" then
+            perform action "AXRaise" of window i
+          end if
+        end repeat
+      end tell
+    end tell`,
+  ]);
+  await sleep(500);
+
+  execFileSync("cliclick", [`c:${screenX},${screenY}`]);
+  console.log("[OK] Start button clicked");
+
+  // Step 7: Wait for huddle to connect
+  console.log("[..] Waiting for huddle to connect...");
+  await sleep(6000);
+
+  // Verify huddle is active
+  const updatedTargets2 = await fetchTargets();
+  const huddleActive = updatedTargets2.some(
+    (t) => t.title?.includes("Huddle:") || t.title?.includes("🎤")
+  );
+
+  if (huddleActive) {
+    console.log("[OK] Huddle is ACTIVE!\n");
+  } else {
+    // Check main window
+    const state = await cdpEval(
+      mainWs,
+      `JSON.stringify({
+        title: document.title,
+        activeAudio: [...document.querySelectorAll('audio')].filter(a => !a.paused).length,
+        huddleChildren: (document.querySelector('div.p-huddle_workspace[role="dialog"]')?.children.length || 0)
+      })`
+    );
+    console.log("[??] Huddle state:", state);
+  }
+
+  // Step 8: Launch huddle server to connect audio bridge
+  console.log("[..] Launching huddle server for audio bridge...\n");
+  const server = spawn("node", [join(__dirname, "huddle-server.mjs")], {
+    stdio: "inherit",
+    env: { ...process.env },
+  });
+
+  server.on("exit", (code) => {
+    console.log(`\n[EXIT] Huddle server exited with code ${code}`);
+    process.exit(code || 0);
+  });
+
+  // Forward signals to the server
+  process.on("SIGINT", () => server.kill("SIGINT"));
+  process.on("SIGTERM", () => server.kill("SIGTERM"));
+}
+
+main().catch((err) => {
+  console.error(`[FATAL] ${err.message}`);
+  process.exit(1);
+});