codegpt-ai 1.27.0 → 1.28.2

package/chat.py

@@ -422,6 +422,8 @@ COMMANDS = {
     "/vote": "All agents vote on a question",
     "/swarm": "Agents collaborate on a task step by step",
     "/team": "Start a team chat with 2 AIs (/team coder reviewer)",
+    "/room": "AI chat room — multiple AIs talk (/room coder reviewer architect)",
+    "/spectate": "Watch AIs chat without you (/spectate claude codex topic)",
     "/github": "GitHub tools (/github repos, issues, prs)",
     "/weather": "Get weather (/weather London)",
     "/open": "Open URL in browser (/open google.com)",
@@ -607,32 +609,27 @@ def ask_permission(action, detail=""):
 
     risk_color = RISK_COLORS.get(risk, "yellow")
     risk_icon = RISK_ICONS.get(risk, "?")
-    compact = is_compact()
 
-    if compact:
-        console.print(Panel(
-            Text.from_markup(
-                f"[{risk_color}]{risk_icon} {risk}[/]\n"
-                f"  {action_desc}\n"
-                + (f"  [dim]{detail[:35]}[/]\n" if detail else "")
-            ),
-            border_style=risk_color.replace("bold ", ""), padding=(0, 1), width=tw(),
-        ))
-    else:
-        console.print(Panel(
-            Text.from_markup(
-                f"[{risk_color}]{risk_icon} Risk: {risk}[/]\n\n"
-                f"  Action: [bright_cyan]{action_desc}[/]\n"
-                + (f"  Detail: [dim]{detail[:60]}[/]\n" if detail else "")
-                + f"\n  [dim](y)es  (n)o  (a)lways allow this[/]"
-            ),
-            title=f"[{risk_color}]Permission[/]",
-            border_style=risk_color.replace("bold ", ""), padding=(0, 2), width=tw(),
-        ))
+    # Risk warnings — explain what could happen
+    risk_warnings = {
+        "CRITICAL": "This can execute code, modify your system, or expose data.",
+        "HIGH": "This accesses external services or modifies important data.",
+        "MEDIUM": "This uses resources or changes session settings.",
+        "LOW": "This is a safe operation with minimal impact.",
+    }
+    warning = risk_warnings.get(risk, "")
+
+    # Clean minimal prompt — like Claude Code
+    console.print()
+    console.print(Text.from_markup(f"  [{risk_color}]{risk_icon} {action_desc}[/]"))
+    if detail:
+        console.print(Text(f"    {detail[:70]}", style="dim"))
+    console.print(Text.from_markup(f"    [{risk_color}]{risk} — {warning}[/]"))
+    console.print()
 
     try:
         answer = prompt(
-            [("class:prompt", " Allow? (y/n/a) > ")],
+            [("class:prompt", "  Allow? (y)es / (n)o / (a)lways > ")],
             style=input_style,
         ).strip().lower()
     except (KeyboardInterrupt, EOFError):
@@ -641,9 +638,9 @@ def ask_permission(action, detail=""):
     if answer in ("a", "always"):
         PERMISSION_ALWAYS_ALLOW.add(action)
         save_permissions()
-        print_sys(f"Always allowed: {action_desc}")
+        console.print(Text(f"  ✓ Always allowed", style="green"))
         return True
-    elif answer in ("y", "yes"):
+    elif answer in ("y", "yes", ""):
         return True
     else:
         print_sys("Denied.")
@@ -1277,13 +1274,38 @@ def _print_err_panel(text):
 
 
 def print_help():
-    table = Table(border_style="bright_black", show_header=True,
-                  header_style="bold bright_cyan", padding=(0, 2))
-    table.add_column("Command", style="bright_cyan", min_width=12)
-    table.add_column("Description", style="dim")
-    for cmd, desc in COMMANDS.items():
-        table.add_row(cmd, desc)
-    console.print(table)
+    # Group commands by category — clean minimal list
+    categories = {
+        "Chat": ["/new", "/save", "/load", "/delete", "/copy", "/regen", "/edit", "/history", "/clear", "/quit"],
+        "Model": ["/model", "/modelinfo", "/params", "/temp", "/think", "/tokens", "/compact"],
+        "AI": ["/agent", "/agents", "/all", "/vote", "/swarm", "/team", "/room", "/spectate", "/dm", "/chat-link"],
+        "Lab": ["/lab", "/chain", "/race", "/prompts"],
+        "Tools": ["/tools", "/bg", "/split", "/grid", "/running", "/killall"],
+        "Connect": ["/connect", "/disconnect", "/server", "/qr", "/scan"],
+        "Files": ["/file", "/run", "/code", "/shell", "/browse", "/open", "/export"],
+        "Memory": ["/mem", "/train", "/pin", "/pins", "/search", "/fork", "/rate", "/tag"],
+        "Profile": ["/profile", "/setname", "/setbio", "/persona", "/personas", "/usage"],
+        "Skills": ["/skill", "/skills", "/auto", "/cron", "/crons"],
+        "Comms": ["/broadcast", "/inbox", "/feed", "/monitor", "/hub"],
+        "System": ["/github", "/weather", "/spotify", "/volume", "/bright", "/sysinfo"],
+        "Security": ["/pin-set", "/pin-remove", "/lock", "/audit", "/security", "/permissions"],
+    }
+
+    on_termux = os.path.exists("/data/data/com.termux")
+
+    for cat, cmds in categories.items():
+        console.print(Text(f"\n  {cat}", style="bold bright_cyan"))
+        for cmd in cmds:
+            desc = COMMANDS.get(cmd, "")
+            if not desc:
+                continue
+            # Hide unsupported tool commands on Termux
+            tool_name = cmd[1:]
+            if on_termux and tool_name in AI_TOOLS and not AI_TOOLS[tool_name].get("termux", True):
+                continue
+            console.print(Text.from_markup(f"    [bright_cyan]{cmd:<16}[/] [dim]{desc}[/]"))
+
+    console.print(Text("\n  Type / to autocomplete. Aliases: /q /n /s /m /h /a /t /f", style="dim"))
     console.print()
 
 
@@ -2412,12 +2434,53 @@ def get_weather(city):
 
 
 def open_url(url):
-    """Open a URL in the default browser."""
+    """Open a URL or search query in the default browser."""
     import webbrowser
-    if not url.startswith("http"):
+
+    # Shortcuts
+    shortcuts = {
+        "google": "https://google.com",
+        "youtube": "https://youtube.com",
+        "github": "https://github.com",
+        "reddit": "https://reddit.com",
+        "twitter": "https://x.com",
+        "x": "https://x.com",
+        "stackoverflow": "https://stackoverflow.com",
+        "npm": "https://npmjs.com",
+        "pypi": "https://pypi.org",
+        "ollama": "https://ollama.com",
+        "claude": "https://claude.ai",
+        "chatgpt": "https://chat.openai.com",
+        "gemini": "https://gemini.google.com",
+    }
+
+    if url.lower() in shortcuts:
+        url = shortcuts[url.lower()]
+    elif "." not in url and ":" not in url:
+        # No dots = search query, not a URL
+        query = url.replace(" ", "+")
+        url = f"https://google.com/search?q={query}"
+    elif not url.startswith("http"):
         url = "https://" + url
-    webbrowser.open(url)
-    print_sys(f"Opened: {url}")
+
+    # Platform-specific browser open
+    try:
+        if os.path.exists("/data/data/com.termux"):
+            # Termux — use termux-open or am start
+            try:
+                subprocess.run(["termux-open-url", url], timeout=5)
+            except FileNotFoundError:
+                subprocess.run(["am", "start", "-a", "android.intent.action.VIEW", "-d", url], timeout=5)
+        elif os.name == "nt":
+            os.startfile(url)
+        elif sys.platform == "darwin":
+            subprocess.run(["open", url], timeout=5)
+        else:
+            webbrowser.open(url)
+        print_sys(f"Opened: {url}")
+    except Exception as e:
+        print_err(f"Cannot open browser: {e}")
+        print_sys(f"URL: {url}")
     audit_log("OPEN_URL", url)
 
 
@@ -3884,6 +3947,150 @@ def team_chat(name1, name2, default_model, system):
     return history
 
 
+# --- Chat Room ---
+
+def chat_room(member_names, default_model, system, user_joins=True, topic=""):
+    """Multi-AI chat room. User can join or spectate."""
+    members = [resolve_team_member(n) for n in member_names]
+
+    names_display = ", ".join(f"[{m['color']}]{m['name']}[/]" for m in members)
+    mode = "Join" if user_joins else "Spectate"
+
+    console.print(Rule(style="bright_green", characters="─"))
+    console.print(Text.from_markup(
+        f"  [bold]Chat Room[/] — {mode} mode\n"
+        f"  Members: {names_display}\n"
+    ))
+    if user_joins:
+        console.print(Text("  Type to talk. @name to address one AI. 'exit' to leave.", style="dim"))
+    else:
+        console.print(Text("  Watching AIs chat. Ctrl+C to stop.", style="dim"))
+    console.print(Rule(style="bright_green", characters="─"))
+    console.print()
+
+    history = []
+
+    if user_joins:
+        # Interactive room — user + multiple AIs
+        while True:
+            try:
+                user_input = prompt(
+                    [("class:prompt", " You ❯ ")],
+                    style=input_style,
+                    history=input_history,
+                ).strip()
+            except (KeyboardInterrupt, EOFError):
+                break
+
+            if not user_input or user_input.lower() in ("exit", "/exit", "quit"):
+                break
+
+            console.print(Text(f"  {user_input}", style="bold white"))
+            console.print()
+            history.append({"speaker": "user", "content": user_input})
+
+            # Check for @mentions
+            mentioned = []
+            for m in members:
+                if f"@{m['name']}" in user_input.lower():
+                    mentioned.append(m)
+
+            # If no mentions, all respond
+            responders = mentioned if mentioned else members
+
+            for member in responders:
+                others = [m['name'] for m in members if m != member] + ["user"]
+                conv = "\n".join(
+                    f"{h['speaker']}: {h['content'][:200]}"
+                    for h in history[-10:]
+                )
+
+                room_prompt = (
+                    f"You are {member['name']} in a group chat with {', '.join(others)}.\n"
+                    f"Chat so far:\n{conv}\n\n"
+                    f"Respond as {member['name']}. Keep it short (2-4 sentences). "
+                    f"React to what was said. Agree, disagree, or add something new. "
+                    f"Don't repeat what others said."
+                )
+
+                try:
+                    resp = requests.post(OLLAMA_URL, json={
+                        "model": member["model"] or default_model,
+                        "messages": [
+                            {"role": "system", "content": member["system"]},
+                            {"role": "user", "content": room_prompt},
+                        ],
+                        "stream": False,
+                    }, timeout=60)
+                    response = resp.json().get("message", {}).get("content", "")
+                except Exception as e:
+                    response = f"(offline)"
+
+                console.print(Text.from_markup(f"  [{member['color']}]{member['name']}[/] {response}"))
+                console.print()
+                history.append({"speaker": member["name"], "content": response})
+                bus_send(member["name"], "codegpt", response[:200], "response")
+
+    else:
+        # Spectate mode — AIs chat with each other, user watches
+        try:
+            # Get initial topic from last arg or default
+            if not topic:
+                topic = "Introduce yourselves and start a technical discussion."
+            if history:
+                topic = history[-1]["content"]
+
+            current_input = topic
+            rounds = 0
+            max_rounds = 12
+
+            while rounds < max_rounds:
+                for member in members:
+                    rounds += 1
+                    if rounds > max_rounds:
+                        break
+
+                    others = [m['name'] for m in members if m != member]
+                    conv = "\n".join(
+                        f"{h['speaker']}: {h['content'][:200]}"
+                        for h in history[-8:]
+                    )
+
+                    room_prompt = (
+                        f"You are {member['name']} in a group chat with {', '.join(others)}.\n"
+                        f"{'Topic: ' + current_input if not history else 'Chat so far:'}\n"
+                        f"{conv}\n\n"
+                        f"Respond as {member['name']}. Keep it short (2-3 sentences). "
+                        f"Build on the conversation. Be opinionated."
+                    )
+
+                    try:
+                        resp = requests.post(OLLAMA_URL, json={
+                            "model": member["model"] or default_model,
+                            "messages": [
+                                {"role": "system", "content": member["system"]},
+                                {"role": "user", "content": room_prompt},
+                            ],
+                            "stream": False,
+                        }, timeout=60)
+                        response = resp.json().get("message", {}).get("content", "")
+                    except Exception as e:
+                        response = "(offline)"
+
+                    console.print(Text.from_markup(f"  [{member['color']}]{member['name']}[/] {response}"))
+                    console.print()
+                    history.append({"speaker": member["name"], "content": response})
+                    time.sleep(0.5)
+
+        except KeyboardInterrupt:
+            pass
+
+    console.print(Rule(style="dim", characters="─"))
+    console.print(Text(f"  Room closed. {len(history)} messages.", style="dim"))
+    console.print()
+    return history
+
+
 # --- Split Screen ---
 
 def get_tool_cmd(name):
@@ -4083,7 +4290,7 @@ def delete_skill(name):
 
 # --- Browser ---
 
-def browse_url(url):
+def browse_url(url, model=None):
     """Fetch a URL, extract text, and summarize it."""
     if not url.startswith("http"):
         url = "https://" + url
@@ -4112,7 +4319,7 @@ def browse_url(url):
         # Ask AI to summarize
         try:
             ai_resp = requests.post(OLLAMA_URL, json={
-                "model": MODEL,
+                "model": model or MODEL,
                 "messages": [
                     {"role": "system", "content": "Summarize this web page content in 3-5 bullet points. Be concise."},
                     {"role": "user", "content": f"URL: {url}\n\nContent:\n{text}"},
@@ -4136,6 +4343,7 @@ def browse_url(url):
 # --- Cron / Scheduled Tasks ---
 
 active_crons = []
+cron_command_queue = []  # Thread-safe command queue for cron execution
 
 
 def add_cron(interval_str, command):
@@ -4161,10 +4369,10 @@ def add_cron(interval_str, command):
             # Check if still active
             if cron_entry not in active_crons:
                 break
-            print_sys(f"[cron] Running: {command}")
-            # Execute as if user typed it
             cron_entry["last_run"] = datetime.now().isoformat()
             cron_entry["runs"] += 1
+            # Queue the command for main loop to execute
+            cron_command_queue.append(command)
 
     cron_entry = {
         "command": command,
@@ -4444,7 +4652,7 @@ def get_input():
 # --- Main ---
 
 def main():
-    global last_ai_response, code_exec_count, OLLAMA_URL, sidebar_enabled
+    global last_ai_response, code_exec_count, OLLAMA_URL, sidebar_enabled, think_mode, temperature
 
     # CLI args mode: python chat.py --ask "question" or python chat.py --cmd "/tools"
     if len(sys.argv) > 1:
@@ -4716,6 +4924,11 @@ def main():
                 audit_log("LOCKED_OUT")
                 break
 
+        # Drain cron command queue
+        while cron_command_queue:
+            cron_cmd = cron_command_queue.pop(0)
+            print_sys(f"[cron] {cron_cmd}")
+
         user_input = get_input()
         if user_input is None:
             cancel_all_reminders()
@@ -4782,10 +4995,10 @@ def main():
                 continue
 
             elif cmd == "/save":
-                if messages and ask_permission("save_chat", "Save conversation"):
-                    save_conversation(messages, model)
-                else:
+                if not messages:
                     print_sys("Nothing to save.")
+                elif ask_permission("save_chat", "Save conversation"):
+                    save_conversation(messages, model)
                 continue
 
             elif cmd == "/load":
@@ -5271,10 +5484,10 @@ def main():
                 elif sub == "clear":
                     mem_clear()
                 elif sub == "inject":
-                    # Inject memories into current conversation as context
+                    # Inject memories as a user-role context message (not system — avoids corrupting conversation)
                     mem_context = get_memory_context()
                     if mem_context:
-                        messages.append({"role": "system", "content": mem_context})
+                        messages.append({"role": "user", "content": f"[Memory context for reference]:\n{mem_context}"})
                         print_sys("Memories injected into context.")
                     else:
                         print_sys("No memories to inject.")
@@ -5656,6 +5869,45 @@ def main():
                     print_sys("Terminal too narrow for sidebar. Widen to 80+ chars.")
                 continue
 
+            elif cmd == "/room":
+                parts = user_input[len("/room "):].strip().split()
+                if len(parts) >= 2:
+                    history = chat_room(parts, model, system, user_joins=True)
+                    for h in history:
+                        if h["speaker"] == "user":
+                            messages.append({"role": "user", "content": h["content"]})
+                        else:
+                            messages.append({"role": "assistant", "content": f"[{h['speaker']}] {h['content']}"})
+                    session_stats["messages"] += len(history)
+                else:
+                    print_sys("Usage: /room coder reviewer architect")
+                    print_sys("       /room claude codex gemini deepseek")
+                    print_sys(f"\nAvailable: {', '.join(list(AI_AGENTS.keys()) + list(TOOL_PERSONAS.keys()))}")
+                continue
+
+            elif cmd == "/spectate":
+                args = user_input[len("/spectate "):].strip().split()
+                if len(args) >= 2:
+                    # Last arg could be a topic
+                    names = args
+                    topic = ""
+                    # Check if last args aren't AI names — treat as topic
+                    all_names = set(AI_AGENTS.keys()) | set(TOOL_PERSONAS.keys())
+                    topic_words = []
+                    while names and names[-1] not in all_names:
+                        topic_words.insert(0, names.pop())
+                    topic = " ".join(topic_words) if topic_words else "Discuss the best programming practices"
+
+                    if len(names) >= 2:
+                        # Inject topic
+                        h = chat_room(names, model, system, user_joins=False, topic=topic)
+                    else:
+                        print_sys("Need at least 2 AIs. Example: /spectate coder reviewer discuss Python")
+                else:
+                    print_sys("Usage: /spectate coder reviewer discuss error handling")
+                    print_sys("       /spectate claude codex gemini debate which is best")
+                continue
+
             elif cmd == "/monitor":
                 # Live updating dashboard — press Ctrl+C to exit
                 console.print(Text("  Live monitor — Ctrl+C to stop\n", style="dim"))
@@ -5979,7 +6231,7 @@ def main():
                         launch_cmd = [tool_bin] + tool.get("default_args", [])
                         if tool_args:
                             launch_cmd.append(tool_args)
-                        subprocess.run(launch_cmd, shell=True, cwd=project_dir, env=tool_env)
+                        subprocess.run(" ".join(launch_cmd), shell=True, cwd=project_dir, env=tool_env)
                     else:
                         tool_sandbox = Path.home() / ".codegpt" / "sandbox" / tool_key
                         tool_sandbox.mkdir(parents=True, exist_ok=True)
@@ -6016,7 +6268,7 @@ def main():
                         launch_cmd = [tool_bin] + tool.get("default_args", [])
                         if tool_args:
                             launch_cmd.append(tool_args)
-                        subprocess.run(launch_cmd, shell=True, cwd=str(tool_sandbox), env=tool_env)
+                        subprocess.run(" ".join(launch_cmd), shell=True, cwd=str(tool_sandbox), env=tool_env)
 
                     print_sys("Back to CodeGPT.")
                     audit_log(f"TOOL_EXIT", tool_key)
@@ -6133,7 +6385,7 @@ def main():
                             print_sys(f"Installed in {elapsed:.1f}s. Launching...")
                             audit_log(f"TOOL_INSTALL", tool_key)
                             launch_cmd = [found_bin] + tool.get("default_args", [])
-                            subprocess.run(launch_cmd, shell=True)
+                            subprocess.run(" ".join(launch_cmd), shell=True)
                             print_sys("Back to CodeGPT.")
                         elif tool_bin in pip_module_map:
                             # Try python -m fallback
@@ -6696,7 +6948,7 @@ def main():
             elif cmd == "/browse":
                 url = user_input[len("/browse "):].strip()
                 if url and ask_permission("open_url", f"Fetch {url}"):
-                    content = browse_url(url)
+                    content = browse_url(url, model=model)
                     if content:
                         messages.append({"role": "user", "content": f"[browsed: {url}]"})
                         messages.append({"role": "assistant", "content": content[:500]})
@@ -6761,21 +7013,23 @@ def main():
                     save_permissions()
                     print_sys("All permissions reset. You'll be asked again.")
                 else:
-                    table = Table(title="Permissions", border_style="yellow",
-                                  title_style="bold yellow", show_header=True, header_style="bold")
-                    table.add_column("Action", style="bright_cyan", width=16)
-                    table.add_column("Description", style="dim")
-                    table.add_column("Risk", width=10)
-                    table.add_column("Status", width=10)
-                    for action, info in RISKY_ACTIONS.items():
-                        if isinstance(info, tuple):
-                            desc, risk = info
-                        else:
-                            desc, risk = info, "MEDIUM"
-                        rc = RISK_COLORS.get(risk, "yellow")
-                        ri = RISK_ICONS.get(risk, "?")
-                        status = "[green]allowed[/]" if action in PERMISSION_ALWAYS_ALLOW else "[yellow]ask[/]"
-                        table.add_row(action, desc, f"[{rc}]{ri} {risk}[/]", status)
+                    console.print(Text("\n  Permissions", style="bold"))
+                    console.print(Rule(style="dim", characters="─"))
+
+                    # Group by risk level
+                    for risk_level in ["CRITICAL", "HIGH", "MEDIUM", "LOW"]:
+                        rc = RISK_COLORS.get(risk_level, "yellow")
+                        ri = RISK_ICONS.get(risk_level, "?")
+                        console.print(Text.from_markup(f"\n  [{rc}]{ri} {risk_level}[/]"))
+                        for action, info in RISKY_ACTIONS.items():
+                            if isinstance(info, tuple):
+                                desc, risk = info
+                            else:
+                                desc, risk = info, "MEDIUM"
+                            if risk != risk_level:
+                                continue
+                            status = "[green]✓ allowed[/]" if action in PERMISSION_ALWAYS_ALLOW else "[dim]ask[/]"
+                            console.print(Text.from_markup(f"    {action:<16} {status}  [dim]{desc}[/]"))
                     console.print(table)
                     console.print(Text("  /permissions reset — revoke all", style="dim"))
                     console.print()
@@ -6809,7 +7063,7 @@ def main():
                         f"  Log entries    [bright_cyan]{audit_count}[/]\n"
                         f"  Log file       [dim]{AUDIT_FILE}[/]\n\n"
                         f"[bold]Storage[/]\n"
-                        f"  Encrypted      [yellow]local XOR[/]\n"
+                        f"  PIN hash       [green]SHA-256[/]\n"
                         f"  Location       [dim]{SECURITY_DIR}[/]\n"
                     ),
                     border_style="bright_cyan",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "codegpt-ai",
-  "version": "1.27.0",
+  "version": "1.28.2",
   "description": "Local AI Assistant Hub — 80+ commands, 29 tools, 8 agents, training, security",
   "author": "ArukuX",
   "license": "MIT",