fix: restore /usage account limits (#958)

agent/account_usage.py

@@ -0,0 +1,326 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from datetime import datetime, timezone
+from typing import Any, Optional
+
+import httpx
+
+from agent.anthropic_adapter import _is_oauth_token, resolve_anthropic_token
+from hermes_cli.auth import _read_codex_tokens, resolve_codex_runtime_credentials
+from hermes_cli.runtime_provider import resolve_runtime_provider
+
+
+def _utc_now() -> datetime:
+    return datetime.now(timezone.utc)
+
+
+@dataclass(frozen=True)
+class AccountUsageWindow:
+    label: str
+    used_percent: Optional[float] = None
+    reset_at: Optional[datetime] = None
+    detail: Optional[str] = None
+
+
+@dataclass(frozen=True)
+class AccountUsageSnapshot:
+    provider: str
+    source: str
+    fetched_at: datetime
+    title: str = "Account limits"
+    plan: Optional[str] = None
+    windows: tuple[AccountUsageWindow, ...] = ()
+    details: tuple[str, ...] = ()
+    unavailable_reason: Optional[str] = None
+
+    @property
+    def available(self) -> bool:
+        return bool(self.windows or self.details) and not self.unavailable_reason
+
+
+def _title_case_slug(value: Optional[str]) -> Optional[str]:
+    cleaned = str(value or "").strip()
+    if not cleaned:
+        return None
+    return cleaned.replace("_", " ").replace("-", " ").title()
+
+
+def _parse_dt(value: Any) -> Optional[datetime]:
+    if value in (None, ""):
+        return None
+    if isinstance(value, (int, float)):
+        return datetime.fromtimestamp(float(value), tz=timezone.utc)
+    if isinstance(value, str):
+        text = value.strip()
+        if not text:
+            return None
+        if text.endswith("Z"):
+            text = text[:-1] + "+00:00"
+        try:
+            dt = datetime.fromisoformat(text)
+            return dt if dt.tzinfo else dt.replace(tzinfo=timezone.utc)
+        except ValueError:
+            return None
+    return None
+
+
+def _format_reset(dt: Optional[datetime]) -> str:
+    if not dt:
+        return "unknown"
+    local_dt = dt.astimezone()
+    delta = dt - _utc_now()
+    total_seconds = int(delta.total_seconds())
+    if total_seconds <= 0:
+        return f"now ({local_dt.strftime('%Y-%m-%d %H:%M %Z')})"
+    hours, rem = divmod(total_seconds, 3600)
+    minutes = rem // 60
+    if hours >= 24:
+        days, hours = divmod(hours, 24)
+        rel = f"in {days}d {hours}h"
+    elif hours > 0:
+        rel = f"in {hours}h {minutes}m"
+    else:
+        rel = f"in {minutes}m"
+    return f"{rel} ({local_dt.strftime('%Y-%m-%d %H:%M %Z')})"
+
+
+def render_account_usage_lines(snapshot: Optional[AccountUsageSnapshot], *, markdown: bool = False) -> list[str]:
+    if not snapshot:
+        return []
+    header = f"📈 {'**' if markdown else ''}{snapshot.title}{'**' if markdown else ''}"
+    lines = [header]
+    if snapshot.plan:
+        lines.append(f"Provider: {snapshot.provider} ({snapshot.plan})")
+    else:
+        lines.append(f"Provider: {snapshot.provider}")
+    for window in snapshot.windows:
+        if window.used_percent is None:
+            base = f"{window.label}: unavailable"
+        else:
+            remaining = max(0, round(100 - float(window.used_percent)))
+            used = max(0, round(float(window.used_percent)))
+            base = f"{window.label}: {remaining}% remaining ({used}% used)"
+        if window.reset_at:
+            base += f" • resets {_format_reset(window.reset_at)}"
+        elif window.detail:
+            base += f" • {window.detail}"
+        lines.append(base)
+    for detail in snapshot.details:
+        lines.append(detail)
+    if snapshot.unavailable_reason:
+        lines.append(f"Unavailable: {snapshot.unavailable_reason}")
+    return lines
+
+
+def _resolve_codex_usage_url(base_url: str) -> str:
+    normalized = (base_url or "").strip().rstrip("/")
+    if not normalized:
+        normalized = "https://chatgpt.com/backend-api/codex"
+    if normalized.endswith("/codex"):
+        normalized = normalized[: -len("/codex")]
+    if "/backend-api" in normalized:
+        return normalized + "/wham/usage"
+    return normalized + "/api/codex/usage"
+
+
+def _fetch_codex_account_usage() -> Optional[AccountUsageSnapshot]:
+    creds = resolve_codex_runtime_credentials(refresh_if_expiring=True)
+    token_data = _read_codex_tokens()
+    tokens = token_data.get("tokens") or {}
+    account_id = str(tokens.get("account_id", "") or "").strip() or None
+    headers = {
+        "Authorization": f"Bearer {creds['api_key']}",
+        "Accept": "application/json",
+        "User-Agent": "codex-cli",
+    }
+    if account_id:
+        headers["ChatGPT-Account-Id"] = account_id
+    with httpx.Client(timeout=15.0) as client:
+        response = client.get(_resolve_codex_usage_url(creds.get("base_url", "")), headers=headers)
+        response.raise_for_status()
+    payload = response.json() or {}
+    rate_limit = payload.get("rate_limit") or {}
+    windows: list[AccountUsageWindow] = []
+    for key, label in (("primary_window", "Session"), ("secondary_window", "Weekly")):
+        window = rate_limit.get(key) or {}
+        used = window.get("used_percent")
+        if used is None:
+            continue
+        windows.append(
+            AccountUsageWindow(
+                label=label,
+                used_percent=float(used),
+                reset_at=_parse_dt(window.get("reset_at")),
+            )
+        )
+    details: list[str] = []
+    credits = payload.get("credits") or {}
+    if credits.get("has_credits"):
+        balance = credits.get("balance")
+        if isinstance(balance, (int, float)):
+            details.append(f"Credits balance: ${float(balance):.2f}")
+        elif credits.get("unlimited"):
+            details.append("Credits balance: unlimited")
+    return AccountUsageSnapshot(
+        provider="openai-codex",
+        source="usage_api",
+        fetched_at=_utc_now(),
+        plan=_title_case_slug(payload.get("plan_type")),
+        windows=tuple(windows),
+        details=tuple(details),
+    )
+
+
+def _fetch_anthropic_account_usage() -> Optional[AccountUsageSnapshot]:
+    token = (resolve_anthropic_token() or "").strip()
+    if not token:
+        return None
+    if not _is_oauth_token(token):
+        return AccountUsageSnapshot(
+            provider="anthropic",
+            source="oauth_usage_api",
+            fetched_at=_utc_now(),
+            unavailable_reason="Anthropic account limits are only available for OAuth-backed Claude accounts.",
+        )
+    headers = {
+        "Authorization": f"Bearer {token}",
+        "Accept": "application/json",
+        "Content-Type": "application/json",
+        "anthropic-beta": "oauth-2025-04-20",
+        "User-Agent": "claude-code/2.1.0",
+    }
+    with httpx.Client(timeout=15.0) as client:
+        response = client.get("https://api.anthropic.com/api/oauth/usage", headers=headers)
+        response.raise_for_status()
+    payload = response.json() or {}
+    windows: list[AccountUsageWindow] = []
+    mapping = (
+        ("five_hour", "Current session"),
+        ("seven_day", "Current week"),
+        ("seven_day_opus", "Opus week"),
+        ("seven_day_sonnet", "Sonnet week"),
+    )
+    for key, label in mapping:
+        window = payload.get(key) or {}
+        util = window.get("utilization")
+        if util is None:
+            continue
+        used = float(util) * 100 if float(util) <= 1 else float(util)
+        windows.append(
+            AccountUsageWindow(
+                label=label,
+                used_percent=used,
+                reset_at=_parse_dt(window.get("resets_at")),
+            )
+        )
+    details: list[str] = []
+    extra = payload.get("extra_usage") or {}
+    if extra.get("is_enabled"):
+        used_credits = extra.get("used_credits")
+        monthly_limit = extra.get("monthly_limit")
+        currency = extra.get("currency") or "USD"
+        if isinstance(used_credits, (int, float)) and isinstance(monthly_limit, (int, float)):
+            details.append(
+                f"Extra usage: {used_credits:.2f} / {monthly_limit:.2f} {currency}"
+            )
+    return AccountUsageSnapshot(
+        provider="anthropic",
+        source="oauth_usage_api",
+        fetched_at=_utc_now(),
+        windows=tuple(windows),
+        details=tuple(details),
+    )
+
+
+def _fetch_openrouter_account_usage(base_url: Optional[str], api_key: Optional[str]) -> Optional[AccountUsageSnapshot]:
+    runtime = resolve_runtime_provider(
+        requested="openrouter",
+        explicit_base_url=base_url,
+        explicit_api_key=api_key,
+    )
+    token = str(runtime.get("api_key", "") or "").strip()
+    if not token:
+        return None
+    normalized = str(runtime.get("base_url", "") or "").rstrip("/")
+    credits_url = f"{normalized}/credits"
+    key_url = f"{normalized}/key"
+    headers = {
+        "Authorization": f"Bearer {token}",
+        "Accept": "application/json",
+    }
+    with httpx.Client(timeout=10.0) as client:
+        credits_resp = client.get(credits_url, headers=headers)
+        credits_resp.raise_for_status()
+        credits = (credits_resp.json() or {}).get("data") or {}
+        try:
+            key_resp = client.get(key_url, headers=headers)
+            key_resp.raise_for_status()
+            key_data = (key_resp.json() or {}).get("data") or {}
+        except Exception:
+            key_data = {}
+    total_credits = float(credits.get("total_credits") or 0.0)
+    total_usage = float(credits.get("total_usage") or 0.0)
+    details = [f"Credits balance: ${max(0.0, total_credits - total_usage):.2f}"]
+    windows: list[AccountUsageWindow] = []
+    limit = key_data.get("limit")
+    limit_remaining = key_data.get("limit_remaining")
+    limit_reset = str(key_data.get("limit_reset") or "").strip()
+    usage = key_data.get("usage")
+    if (
+        isinstance(limit, (int, float))
+        and float(limit) > 0
+        and isinstance(limit_remaining, (int, float))
+        and 0 <= float(limit_remaining) <= float(limit)
+    ):
+        limit_value = float(limit)
+        remaining_value = float(limit_remaining)
+        used_percent = ((limit_value - remaining_value) / limit_value) * 100
+        detail_parts = [f"${remaining_value:.2f} of ${limit_value:.2f} remaining"]
+        if limit_reset:
+            detail_parts.append(f"resets {limit_reset}")
+        windows.append(
+            AccountUsageWindow(
+                label="API key quota",
+                used_percent=used_percent,
+                detail=" • ".join(detail_parts),
+            )
+        )
+    if isinstance(usage, (int, float)):
+        usage_parts = [f"API key usage: ${float(usage):.2f} total"]
+        for value, label in (
+            (key_data.get("usage_daily"), "today"),
+            (key_data.get("usage_weekly"), "this week"),
+            (key_data.get("usage_monthly"), "this month"),
+        ):
+            if isinstance(value, (int, float)) and float(value) > 0:
+                usage_parts.append(f"${float(value):.2f} {label}")
+        details.append(" • ".join(usage_parts))
+    return AccountUsageSnapshot(
+        provider="openrouter",
+        source="credits_api",
+        fetched_at=_utc_now(),
+        windows=tuple(windows),
+        details=tuple(details),
+    )
+
+
+def fetch_account_usage(
+    provider: Optional[str],
+    *,
+    base_url: Optional[str] = None,
+    api_key: Optional[str] = None,
+) -> Optional[AccountUsageSnapshot]:
+    normalized = str(provider or "").strip().lower()
+    if normalized in {"", "auto", "custom"}:
+        return None
+    try:
+        if normalized == "openai-codex":
+            return _fetch_codex_account_usage()
+        if normalized == "anthropic":
+            return _fetch_anthropic_account_usage()
+        if normalized == "openrouter":
+            return _fetch_openrouter_account_usage(base_url, api_key)
+    except Exception:
+        return None
+    return None

cli.py

@@ -13,6 +13,7 @@ Usage:
     python cli.py --list-tools             # List available tools and exit
 """
 
+import concurrent.futures
 import logging
 import os
 import shutil
@@ -63,6 +64,7 @@ from agent.usage_pricing import (
     format_duration_compact,
     format_token_count_compact,
 )
+from agent.account_usage import fetch_account_usage, render_account_usage_lines
 from hermes_cli.banner import _format_context_length, format_banner_version_label
 
 _COMMAND_SPINNER_FRAMES = ("⠋", "⠙", "⠹", "⠸", "⠼", "⠴", "⠦", "⠧", "⠇", "⠏")
@@ -6471,6 +6473,29 @@ class HermesCLI:
         if cost_result.status == "unknown":
             print(f"  Note:             Pricing unknown for {agent.model}")
 
+        # Account limits -- fetched off-thread with a hard timeout so slow
+        # provider APIs don't hang the prompt.
+        provider = getattr(agent, "provider", None) or getattr(self, "provider", None)
+        base_url = getattr(agent, "base_url", None) or getattr(self, "base_url", None)
+        api_key = getattr(agent, "api_key", None) or getattr(self, "api_key", None)
+        account_snapshot = None
+        if provider:
+            with concurrent.futures.ThreadPoolExecutor(max_workers=1) as _pool:
+                try:
+                    account_snapshot = _pool.submit(
+                        fetch_account_usage,
+                        provider,
+                        base_url=base_url,
+                        api_key=api_key,
+                    ).result(timeout=10.0)
+                except (concurrent.futures.TimeoutError, Exception):
+                    account_snapshot = None
+        account_lines = [f"  {line}" for line in render_account_usage_lines(account_snapshot)]
+        if account_lines:
+            print()
+            for line in account_lines:
+                print(line)
+
         if self.verbose:
             logging.getLogger().setLevel(logging.DEBUG)
             for noisy in ('openai', 'openai._base_client', 'httpx', 'httpcore', 'asyncio', 'hpack', 'grpc', 'modal'):

gateway/run.py

@@ -28,6 +28,8 @@ from pathlib import Path
 from datetime import datetime
 from typing import Dict, Optional, Any, List
 
+from agent.account_usage import fetch_account_usage, render_account_usage_lines
+
 # ---------------------------------------------------------------------------
 # SSL certificate auto-detection for NixOS and other non-standard systems.
 # Must run BEFORE any HTTP library (discord, aiohttp, etc.) is imported.
@@ -6481,6 +6483,38 @@ class GatewayRunner:
                     if cached:
                         agent = cached[0]
 
+        # Resolve provider/base_url/api_key for the account-usage fetch.
+        # Prefer the live agent; fall back to persisted billing data on the
+        # SessionDB row so `/usage` still returns account info between turns
+        # when no agent is resident.
+        provider = getattr(agent, "provider", None) if agent and agent is not _AGENT_PENDING_SENTINEL else None
+        base_url = getattr(agent, "base_url", None) if agent and agent is not _AGENT_PENDING_SENTINEL else None
+        api_key = getattr(agent, "api_key", None) if agent and agent is not _AGENT_PENDING_SENTINEL else None
+        if not provider and getattr(self, "_session_db", None) is not None:
+            try:
+                _entry_for_billing = self.session_store.get_or_create_session(source)
+                persisted = self._session_db.get_session(_entry_for_billing.session_id) or {}
+            except Exception:
+                persisted = {}
+            provider = provider or persisted.get("billing_provider")
+            base_url = base_url or persisted.get("billing_base_url")
+
+        # Fetch account usage off the event loop so slow provider APIs don't
+        # block the gateway. Failures are non-fatal -- account_lines stays [].
+        account_lines: list[str] = []
+        if provider:
+            try:
+                account_snapshot = await asyncio.to_thread(
+                    fetch_account_usage,
+                    provider,
+                    base_url=base_url,
+                    api_key=api_key,
+                )
+            except Exception:
+                account_snapshot = None
+            if account_snapshot:
+                account_lines = render_account_usage_lines(account_snapshot, markdown=True)
+
         if agent and hasattr(agent, "session_total_tokens") and agent.session_api_calls > 0:
             lines = []
 
@@ -6538,6 +6572,10 @@ class GatewayRunner:
             if ctx.compression_count:
                 lines.append(f"Compressions: {ctx.compression_count}")
 
+            if account_lines:
+                lines.append("")
+                lines.extend(account_lines)
+
             return "\n".join(lines)
 
         # No agent at all -- check session history for a rough count
@@ -6547,12 +6585,18 @@ class GatewayRunner:
             from agent.model_metadata import estimate_messages_tokens_rough
             msgs = [m for m in history if m.get("role") in ("user", "assistant") and m.get("content")]
             approx = estimate_messages_tokens_rough(msgs)
-            return (
-                f"📊 **Session Info**\n"
-                f"Messages: {len(msgs)}\n"
-                f"Estimated context: ~{approx:,} tokens\n"
-                f"_(Detailed usage available after the first agent response)_"
-            )
+            lines = [
+                "📊 **Session Info**",
+                f"Messages: {len(msgs)}",
+                f"Estimated context: ~{approx:,} tokens",
+                "_(Detailed usage available after the first agent response)_",
+            ]
+            if account_lines:
+                lines.append("")
+                lines.extend(account_lines)
+            return "\n".join(lines)
+        if account_lines:
+            return "\n".join(account_lines)
         return "No usage data available for this session."
 
     async def _handle_insights_command(self, event: MessageEvent) -> str:

hermes_cli/config.py

@@ -523,7 +523,7 @@ DEFAULT_CONFIG = {
     
     # Text-to-speech configuration
     "tts": {
-        "provider": "edge",  # "edge" (free) | "elevenlabs" (premium) | "openai" | "minimax" | "mistral" | "neutts" (local) | "kittentts" (local)
+        "provider": "edge",  # "edge" (free) | "elevenlabs" (premium) | "openai" | "minimax" | "mistral" | "neutts" (local)
         "edge": {
             "voice": "en-US-AriaNeural",
             # Popular: AriaNeural, JennyNeural, AndrewNeural, BrianNeural, SoniaNeural
@@ -547,12 +547,6 @@ DEFAULT_CONFIG = {
             "model": "neuphonic/neutts-air-q4-gguf",  # HuggingFace model repo
             "device": "cpu",  # cpu, cuda, or mps
         },
-        "kittentts": {
-            "model": "KittenML/kitten-tts-nano-0.8-int8",  # 25MB int8 default
-            "voice": "Jasper",  # Jasper, Bella, Luna, Bruno, Rosie, Hugo, Kiki, Leo
-            "speed": 1.0,
-            "clean_text": True,
-        },
     },
     
     "stt": {

hermes_cli/setup.py

@@ -443,16 +443,6 @@ def _print_setup_summary(config: dict, hermes_home):
             tool_status.append(("Text-to-Speech (NeuTTS local)", True, None))
         else:
             tool_status.append(("Text-to-Speech (NeuTTS — not installed)", False, "run 'hermes setup tts'"))
-    elif tts_provider == "kittentts":
-        try:
-            import importlib.util
-            kittentts_ok = importlib.util.find_spec("kittentts") is not None
-        except Exception:
-            kittentts_ok = False
-        if kittentts_ok:
-            tool_status.append(("Text-to-Speech (KittenTTS local)", True, None))
-        else:
-            tool_status.append(("Text-to-Speech (KittenTTS — not installed)", False, "run 'hermes setup tts'"))
     else:
         tool_status.append(("Text-to-Speech (Edge TTS)", True, None))
 
@@ -901,7 +891,6 @@ def _install_neutts_deps() -> bool:
                 return False
         else:
             print_warning("espeak-ng is required for NeuTTS. Install it manually before using NeuTTS.")
-            return False
 
     # Install neutts Python package
     print()
@@ -921,34 +910,8 @@ def _install_neutts_deps() -> bool:
         return False
 
 
-def _install_kittentts_deps() -> bool:
-    """Install KittenTTS dependencies with user approval. Returns True on success."""
-    import subprocess
-    import sys
-
-    wheel_url = (
-        "https://github.com/KittenML/KittenTTS/releases/download/"
-        "0.8.1/kittentts-0.8.1-py3-none-any.whl"
-    )
-    print()
-    print_info("Installing kittentts Python package (~25-80MB model downloaded on first use)...")
-    print()
-    try:
-        subprocess.run(
-            [sys.executable, "-m", "pip", "install", "-U", wheel_url, "soundfile", "--quiet"],
-            check=True, timeout=300,
-        )
-        print_success("kittentts installed successfully")
-        return True
-    except (subprocess.CalledProcessError, subprocess.TimeoutExpired) as e:
-        print_error(f"Failed to install kittentts: {e}")
-        print_info(f"Try manually: python -m pip install -U '{wheel_url}' soundfile")
-        return False
-
-
 def _setup_tts_provider(config: dict):
-    """Interactive TTS provider selection with install flow for local providers."""
-
+    """Interactive TTS provider selection with install flow for NeuTTS."""
     tts_config = config.get("tts", {})
     current_provider = tts_config.get("provider", "edge")
     subscription_features = get_nous_subscription_features(config)
@@ -960,7 +923,6 @@ def _setup_tts_provider(config: dict):
         "minimax": "MiniMax TTS",
         "mistral": "Mistral Voxtral TTS",
         "neutts": "NeuTTS",
-        "kittentts": "KittenTTS",
     }
     current_label = provider_labels.get(current_provider, current_provider)
 
@@ -982,10 +944,9 @@ def _setup_tts_provider(config: dict):
             "MiniMax TTS (high quality with voice cloning, needs API key)",
             "Mistral Voxtral TTS (multilingual, native Opus, needs API key)",
             "NeuTTS (local on-device, free, ~300MB model download)",
-            "KittenTTS (local on-device, free, lightweight ~25-80MB ONNX)",
         ]
     )
-    providers.extend(["edge", "elevenlabs", "openai", "minimax", "mistral", "neutts", "kittentts"])
+    providers.extend(["edge", "elevenlabs", "openai", "minimax", "mistral", "neutts"])
     choices.append(f"Keep current ({current_label})")
     keep_current_idx = len(choices) - 1
     idx = prompt_choice("Select TTS provider:", choices, keep_current_idx)
@@ -1027,28 +988,6 @@ def _setup_tts_provider(config: dict):
                 print_info("Skipping install. Set tts.provider to 'neutts' after installing manually.")
                 selected = "edge"
 
-    elif selected == "kittentts":
-        try:
-            import importlib.util
-            already_installed = importlib.util.find_spec("kittentts") is not None
-        except Exception:
-            already_installed = False
-
-        if already_installed:
-            print_success("KittenTTS is already installed")
-        else:
-            print()
-            print_info("KittenTTS is lightweight (~25-80MB, CPU-only, no API key required).")
-            print_info("Voices: Jasper, Bella, Luna, Bruno, Rosie, Hugo, Kiki, Leo")
-            print()
-            if prompt_yes_no("Install KittenTTS now?", True):
-                if not _install_kittentts_deps():
-                    print_warning("KittenTTS installation incomplete. Falling back to Edge TTS.")
-                    selected = "edge"
-            else:
-                print_info("Skipping install. Set tts.provider to 'kittentts' after installing manually.")
-                selected = "edge"
-
     elif selected == "elevenlabs":
         existing = get_env_value("ELEVENLABS_API_KEY")
         if not existing:

hermes_cli/tools_config.py

@@ -164,14 +164,6 @@ TOOL_CATEGORIES = {
                 ],
                 "tts_provider": "mistral",
             },
-            {
-                "name": "KittenTTS",
-                "badge": "local · free",
-                "tag": "Lightweight local ONNX TTS (~25MB), no API key",
-                "env_vars": [],
-                "tts_provider": "kittentts",
-                "post_setup": "kittentts",
-            },
         ],
     },
     "web": {
@@ -411,36 +403,6 @@ def _run_post_setup(post_setup_key: str):
             _print_warning("    Node.js not found. Install Camofox via Docker:")
             _print_info("      docker run -p 9377:9377 -e CAMOFOX_PORT=9377 jo-inc/camofox-browser")
 
-    elif post_setup_key == "kittentts":
-        try:
-            __import__("kittentts")
-            _print_success("    kittentts is already installed")
-            return
-        except ImportError:
-            pass
-        import subprocess
-        _print_info("    Installing kittentts (~25-80MB model, CPU-only)...")
-        wheel_url = (
-            "https://github.com/KittenML/KittenTTS/releases/download/"
-            "0.8.1/kittentts-0.8.1-py3-none-any.whl"
-        )
-        try:
-            result = subprocess.run(
-                [sys.executable, "-m", "pip", "install", "-U", wheel_url, "soundfile", "--quiet"],
-                capture_output=True, text=True, timeout=300,
-            )
-            if result.returncode == 0:
-                _print_success("    kittentts installed")
-                _print_info("    Voices: Jasper, Bella, Luna, Bruno, Rosie, Hugo, Kiki, Leo")
-                _print_info("    Models: KittenML/kitten-tts-nano-0.8-int8 (25MB), micro (41MB), mini (80MB)")
-            else:
-                _print_warning("    kittentts install failed:")
-                _print_info(f"      {result.stderr.strip()[:300]}")
-                _print_info(f"    Run manually: python -m pip install -U '{wheel_url}' soundfile")
-        except subprocess.TimeoutExpired:
-            _print_warning("    kittentts install timed out (>5min)")
-            _print_info(f"    Run manually: python -m pip install -U '{wheel_url}' soundfile")
-
     elif post_setup_key == "rl_training":
         try:
             __import__("tinker_atropos")

tests/gateway/test_usage_command.py

@@ -175,3 +175,79 @@ class TestUsageCachedAgent:
             result = await runner._handle_usage_command(event)
 
         assert "Cost: included" in result
+
+
+class TestUsageAccountSection:
+    """Account-limits section appended to /usage output."""
+
+    @pytest.mark.asyncio
+    async def test_usage_command_includes_account_section(self, monkeypatch):
+        agent = _make_mock_agent(provider="openai-codex")
+        agent.base_url = "https://chatgpt.com/backend-api/codex"
+        agent.api_key = "unused"
+        runner = _make_runner(SK, cached_agent=agent)
+        event = MagicMock()
+
+        monkeypatch.setattr(
+            "gateway.run.fetch_account_usage",
+            lambda provider, base_url=None, api_key=None: object(),
+        )
+        monkeypatch.setattr(
+            "gateway.run.render_account_usage_lines",
+            lambda snapshot, markdown=False: [
+                "📈 **Account limits**",
+                "Provider: openai-codex (Pro)",
+                "Session: 85% remaining (15% used)",
+            ],
+        )
+        with patch("agent.rate_limit_tracker.format_rate_limit_compact", return_value="RPM: 50/60"), \
+             patch("agent.usage_pricing.estimate_usage_cost") as mock_cost:
+            mock_cost.return_value = MagicMock(amount_usd=None, status="included")
+            result = await runner._handle_usage_command(event)
+
+        assert "📊 **Session Token Usage**" in result
+        assert "📈 **Account limits**" in result
+        assert "Provider: openai-codex (Pro)" in result
+
+    @pytest.mark.asyncio
+    async def test_usage_command_uses_persisted_provider_when_agent_not_running(self, monkeypatch):
+        runner = _make_runner(SK)
+        runner._session_db = MagicMock()
+        runner._session_db.get_session.return_value = {
+            "billing_provider": "openai-codex",
+            "billing_base_url": "https://chatgpt.com/backend-api/codex",
+        }
+        session_entry = MagicMock()
+        session_entry.session_id = "sess-1"
+        runner.session_store.get_or_create_session.return_value = session_entry
+        runner.session_store.load_transcript.return_value = [
+            {"role": "user", "content": "earlier"},
+        ]
+
+        calls = {}
+
+        async def _fake_to_thread(fn, *args, **kwargs):
+            calls["args"] = args
+            calls["kwargs"] = kwargs
+            return fn(*args, **kwargs)
+
+        monkeypatch.setattr("gateway.run.asyncio.to_thread", _fake_to_thread)
+        monkeypatch.setattr(
+            "gateway.run.fetch_account_usage",
+            lambda provider, base_url=None, api_key=None: object(),
+        )
+        monkeypatch.setattr(
+            "gateway.run.render_account_usage_lines",
+            lambda snapshot, markdown=False: [
+                "📈 **Account limits**",
+                "Provider: openai-codex (Pro)",
+            ],
+        )
+
+        event = MagicMock()
+        result = await runner._handle_usage_command(event)
+
+        assert calls["args"] == ("openai-codex",)
+        assert calls["kwargs"]["base_url"] == "https://chatgpt.com/backend-api/codex"
+        assert "📊 **Session Info**" in result
+        assert "📈 **Account limits**" in result

tests/test_account_usage.py

@@ -0,0 +1,203 @@
+from datetime import datetime, timezone
+
+from agent.account_usage import (
+    AccountUsageSnapshot,
+    AccountUsageWindow,
+    fetch_account_usage,
+    render_account_usage_lines,
+)
+
+
+class _Response:
+    def __init__(self, payload, status_code=200):
+        self._payload = payload
+        self.status_code = status_code
+
+    def raise_for_status(self):
+        if self.status_code >= 400:
+            raise RuntimeError(f"HTTP {self.status_code}")
+
+    def json(self):
+        return self._payload
+
+
+class _Client:
+    def __init__(self, payload):
+        self._payload = payload
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc, tb):
+        return False
+
+    def get(self, url, headers=None):
+        return _Response(self._payload)
+
+
+class _RoutingClient:
+    def __init__(self, payloads):
+        self._payloads = payloads
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc, tb):
+        return False
+
+    def get(self, url, headers=None):
+        return _Response(self._payloads[url])
+
+
+def test_fetch_account_usage_codex(monkeypatch):
+    monkeypatch.setattr(
+        "agent.account_usage.resolve_codex_runtime_credentials",
+        lambda refresh_if_expiring=True: {
+            "provider": "openai-codex",
+            "base_url": "https://chatgpt.com/backend-api/codex",
+            "api_key": "***",
+        },
+    )
+    monkeypatch.setattr(
+        "agent.account_usage._read_codex_tokens",
+        lambda: {"tokens": {"account_id": "acct_123"}},
+    )
+    monkeypatch.setattr(
+        "agent.account_usage.httpx.Client",
+        lambda timeout=15.0: _Client(
+            {
+                "plan_type": "pro",
+                "rate_limit": {
+                    "primary_window": {
+                        "used_percent": 15,
+                        "reset_at": 1_900_000_000,
+                        "limit_window_seconds": 18000,
+                    },
+                    "secondary_window": {
+                        "used_percent": 40,
+                        "reset_at": 1_900_500_000,
+                        "limit_window_seconds": 604800,
+                    },
+                },
+                "credits": {"has_credits": True, "balance": 12.5},
+            }
+        ),
+    )
+
+    snapshot = fetch_account_usage("openai-codex")
+
+    assert snapshot is not None
+    assert snapshot.plan == "Pro"
+    assert len(snapshot.windows) == 2
+    assert snapshot.windows[0].label == "Session"
+    assert snapshot.windows[0].used_percent == 15.0
+    assert snapshot.windows[0].reset_at == datetime.fromtimestamp(1_900_000_000, tz=timezone.utc)
+    assert "Credits balance: $12.50" in snapshot.details
+
+
+def test_render_account_usage_lines_includes_reset_and_provider():
+    snapshot = AccountUsageSnapshot(
+        provider="openai-codex",
+        source="usage_api",
+        fetched_at=datetime.now(timezone.utc),
+        plan="Pro",
+        windows=(
+            AccountUsageWindow(
+                label="Session",
+                used_percent=25,
+                reset_at=datetime.now(timezone.utc),
+            ),
+        ),
+        details=("Credits balance: $9.99",),
+    )
+    lines = render_account_usage_lines(snapshot)
+
+    assert lines[0] == "📈 Account limits"
+    assert "openai-codex (Pro)" in lines[1]
+    assert "Session: 75% remaining (25% used)" in lines[2]
+    assert "Credits balance: $9.99" in lines[3]
+
+
+def test_fetch_account_usage_openrouter_uses_limit_remaining_and_ignores_deprecated_rate_limit(monkeypatch):
+    monkeypatch.setattr(
+        "agent.account_usage.resolve_runtime_provider",
+        lambda requested, explicit_base_url=None, explicit_api_key=None: {
+            "provider": "openrouter",
+            "base_url": "https://openrouter.ai/api/v1",
+            "api_key": "***",
+        },
+    )
+    monkeypatch.setattr(
+        "agent.account_usage.httpx.Client",
+        lambda timeout=10.0: _RoutingClient(
+            {
+                "https://openrouter.ai/api/v1/credits": {
+                    "data": {"total_credits": 300.0, "total_usage": 10.92}
+                },
+                "https://openrouter.ai/api/v1/key": {
+                    "data": {
+                        "limit": 100.0,
+                        "limit_remaining": 70.0,
+                        "limit_reset": "monthly",
+                        "usage": 12.5,
+                        "usage_daily": 0.5,
+                        "usage_weekly": 2.0,
+                        "usage_monthly": 8.0,
+                        "rate_limit": {"requests": -1, "interval": "10s"},
+                    }
+                },
+            }
+        ),
+    )
+
+    snapshot = fetch_account_usage("openrouter")
+
+    assert snapshot is not None
+    assert snapshot.windows == (
+        AccountUsageWindow(
+            label="API key quota",
+            used_percent=30.0,
+            detail="$70.00 of $100.00 remaining • resets monthly",
+        ),
+    )
+    assert "Credits balance: $289.08" in snapshot.details
+    assert "API key usage: $12.50 total • $0.50 today • $2.00 this week • $8.00 this month" in snapshot.details
+    assert all("-1 requests / 10s" not in line for line in render_account_usage_lines(snapshot))
+
+
+def test_fetch_account_usage_openrouter_omits_quota_window_when_key_has_no_limit(monkeypatch):
+    monkeypatch.setattr(
+        "agent.account_usage.resolve_runtime_provider",
+        lambda requested, explicit_base_url=None, explicit_api_key=None: {
+            "provider": "openrouter",
+            "base_url": "https://openrouter.ai/api/v1",
+            "api_key": "***",
+        },
+    )
+    monkeypatch.setattr(
+        "agent.account_usage.httpx.Client",
+        lambda timeout=10.0: _RoutingClient(
+            {
+                "https://openrouter.ai/api/v1/credits": {
+                    "data": {"total_credits": 100.0, "total_usage": 25.5}
+                },
+                "https://openrouter.ai/api/v1/key": {
+                    "data": {
+                        "limit": None,
+                        "limit_remaining": None,
+                        "usage": 25.5,
+                        "usage_daily": 1.25,
+                        "usage_weekly": 4.5,
+                        "usage_monthly": 18.0,
+                    }
+                },
+            }
+        ),
+    )
+
+    snapshot = fetch_account_usage("openrouter")
+
+    assert snapshot is not None
+    assert snapshot.windows == ()
+    assert "Credits balance: $74.50" in snapshot.details
+    assert "API key usage: $25.50 total • $1.25 today • $4.50 this week • $18.00 this month" in snapshot.details

tests/tools/test_tts_kittentts.py

@@ -1,236 +0,0 @@
-"""Tests for the KittenTTS local provider in tools/tts_tool.py."""
-
-import json
-from unittest.mock import MagicMock, patch
-
-import numpy as np
-import pytest
-
-
-@pytest.fixture(autouse=True)
-def clean_env(monkeypatch):
-    for key in ("HERMES_SESSION_PLATFORM",):
-        monkeypatch.delenv(key, raising=False)
-
-
-@pytest.fixture(autouse=True)
-def clear_kittentts_cache():
-    """Reset the module-level model cache between tests."""
-    from tools import tts_tool as _tt
-    _tt._kittentts_model_cache.clear()
-    yield
-    _tt._kittentts_model_cache.clear()
-
-
-@pytest.fixture
-def mock_kittentts_module():
-    """Inject a fake kittentts + soundfile module that return stub objects."""
-    fake_model = MagicMock()
-    # 24kHz float32 PCM at ~2s of silence
-    fake_model.generate.return_value = np.zeros(48000, dtype=np.float32)
-    fake_cls = MagicMock(return_value=fake_model)
-    fake_kittentts = MagicMock()
-    fake_kittentts.KittenTTS = fake_cls
-
-    # Stub soundfile — the real package isn't installed in CI venv, and
-    # _generate_kittentts does `import soundfile as sf` at runtime.
-    fake_sf = MagicMock()
-
-    def _fake_write(path, audio, samplerate):
-        # Emulate writing a real file so downstream path checks succeed.
-        import pathlib
-
-        pathlib.Path(path).write_bytes(b"RIFF\x00\x00\x00\x00WAVEfmt fake")
-
-    fake_sf.write = _fake_write
-
-    with patch.dict(
-        "sys.modules",
-        {"kittentts": fake_kittentts, "soundfile": fake_sf},
-    ):
-        yield fake_model, fake_cls
-
-
-class TestGenerateKittenTts:
-    def test_successful_wav_generation(self, tmp_path, mock_kittentts_module):
-        from tools.tts_tool import _generate_kittentts
-
-        fake_model, fake_cls = mock_kittentts_module
-        output_path = str(tmp_path / "test.wav")
-        result = _generate_kittentts("Hello world", output_path, {})
-
-        assert result == output_path
-        assert (tmp_path / "test.wav").exists()
-        fake_cls.assert_called_once()
-        fake_model.generate.assert_called_once()
-
-    def test_config_passes_voice_speed_cleantext(self, tmp_path, mock_kittentts_module):
-        from tools.tts_tool import _generate_kittentts
-
-        fake_model, _ = mock_kittentts_module
-        config = {
-            "kittentts": {
-                "model": "KittenML/kitten-tts-mini-0.8",
-                "voice": "Luna",
-                "speed": 1.25,
-                "clean_text": False,
-            }
-        }
-        _generate_kittentts("Hi there", str(tmp_path / "out.wav"), config)
-
-        call_kwargs = fake_model.generate.call_args.kwargs
-        assert call_kwargs["voice"] == "Luna"
-        assert call_kwargs["speed"] == 1.25
-        assert call_kwargs["clean_text"] is False
-
-    def test_default_model_and_voice(self, tmp_path, mock_kittentts_module):
-        from tools.tts_tool import (
-            DEFAULT_KITTENTTS_MODEL,
-            DEFAULT_KITTENTTS_VOICE,
-            _generate_kittentts,
-        )
-
-        fake_model, fake_cls = mock_kittentts_module
-        _generate_kittentts("Hi", str(tmp_path / "out.wav"), {})
-
-        fake_cls.assert_called_once_with(DEFAULT_KITTENTTS_MODEL)
-        assert fake_model.generate.call_args.kwargs["voice"] == DEFAULT_KITTENTTS_VOICE
-
-    def test_model_is_cached_across_calls(self, tmp_path, mock_kittentts_module):
-        from tools.tts_tool import _generate_kittentts
-
-        _, fake_cls = mock_kittentts_module
-        _generate_kittentts("One", str(tmp_path / "a.wav"), {})
-        _generate_kittentts("Two", str(tmp_path / "b.wav"), {})
-
-        # Same model name → class instantiated exactly once
-        assert fake_cls.call_count == 1
-
-    def test_different_models_are_cached_separately(self, tmp_path, mock_kittentts_module):
-        from tools.tts_tool import _generate_kittentts
-
-        _, fake_cls = mock_kittentts_module
-        _generate_kittentts(
-            "A",
-            str(tmp_path / "a.wav"),
-            {"kittentts": {"model": "KittenML/kitten-tts-nano-0.8-int8"}},
-        )
-        _generate_kittentts(
-            "B",
-            str(tmp_path / "b.wav"),
-            {"kittentts": {"model": "KittenML/kitten-tts-mini-0.8"}},
-        )
-
-        assert fake_cls.call_count == 2
-
-    def test_non_wav_extension_triggers_ffmpeg_conversion(
-        self, tmp_path, mock_kittentts_module, monkeypatch
-    ):
-        """Non-.wav output path causes WAV → target ffmpeg conversion."""
-        from tools import tts_tool as _tt
-
-        calls = []
-
-        def fake_shutil_which(cmd):
-            return "/usr/bin/ffmpeg" if cmd == "ffmpeg" else None
-
-        def fake_run(cmd, check=False, timeout=None, **kw):
-            calls.append(cmd)
-            # Emulate ffmpeg writing the output file
-            import pathlib
-
-            out_path = cmd[-1]
-            pathlib.Path(out_path).write_bytes(b"fake-mp3-data")
-            return MagicMock(returncode=0)
-
-        monkeypatch.setattr(_tt.shutil, "which", fake_shutil_which)
-        monkeypatch.setattr(_tt.subprocess, "run", fake_run)
-
-        output_path = str(tmp_path / "test.mp3")
-        result = _tt._generate_kittentts("Hi", output_path, {})
-
-        assert result == output_path
-        assert len(calls) == 1
-        assert calls[0][0] == "/usr/bin/ffmpeg"
-
-    def test_missing_kittentts_raises_import_error(self, tmp_path, monkeypatch):
-        """When kittentts package is not installed, _import_kittentts raises."""
-        import sys
-
-        monkeypatch.setitem(sys.modules, "kittentts", None)
-        from tools.tts_tool import _generate_kittentts
-
-        with pytest.raises((ImportError, TypeError)):
-            _generate_kittentts("Hi", str(tmp_path / "out.wav"), {})
-
-
-class TestCheckKittenttsAvailable:
-    def test_reports_available_when_package_present(self, monkeypatch):
-        import importlib.util
-        from tools.tts_tool import _check_kittentts_available
-
-        fake_spec = MagicMock()
-        monkeypatch.setattr(
-            importlib.util,
-            "find_spec",
-            lambda name: fake_spec if name == "kittentts" else None,
-        )
-        assert _check_kittentts_available() is True
-
-    def test_reports_unavailable_when_package_missing(self, monkeypatch):
-        import importlib.util
-        from tools.tts_tool import _check_kittentts_available
-
-        monkeypatch.setattr(importlib.util, "find_spec", lambda name: None)
-        assert _check_kittentts_available() is False
-
-
-class TestDispatcherBranch:
-    def test_kittentts_not_installed_returns_helpful_error(self, monkeypatch, tmp_path):
-        """When provider=kittentts but package missing, return JSON error with setup hint."""
-        import sys
-
-        monkeypatch.setitem(sys.modules, "kittentts", None)
-        monkeypatch.setenv("HERMES_HOME", str(tmp_path))
-
-        from tools.tts_tool import text_to_speech_tool
-
-        # Write a config telling it to use kittentts
-        import yaml
-
-        (tmp_path / "config.yaml").write_text(
-            yaml.safe_dump({"tts": {"provider": "kittentts"}})
-        )
-
-        result = json.loads(text_to_speech_tool(text="Hello"))
-        assert result["success"] is False
-        assert "kittentts" in result["error"].lower()
-        assert "hermes setup tts" in result["error"].lower()
-
-    def test_non_telegram_explicit_wav_path_is_preserved(
-        self, monkeypatch, tmp_path, mock_kittentts_module
-    ):
-        """Explicit WAV outputs should stay WAV outside Telegram sessions."""
-        import yaml
-        from tools import tts_tool as _tt
-
-        monkeypatch.setenv("HERMES_HOME", str(tmp_path))
-        (tmp_path / "config.yaml").write_text(
-            yaml.safe_dump({"tts": {"provider": "kittentts"}})
-        )
-
-        def fail_convert(_path):
-            raise AssertionError("_convert_to_opus should not run outside Telegram")
-
-        monkeypatch.setattr(_tt, "_convert_to_opus", fail_convert)
-
-        result = json.loads(
-            _tt.text_to_speech_tool(
-                text="Hello from KittenTTS",
-                output_path=str(tmp_path / "out.wav"),
-            )
-        )
-
-        assert result["success"] is True
-        assert result["file_path"] == str(tmp_path / "out.wav")
-        assert (tmp_path / "out.wav").exists()

tools/tts_tool.py

@@ -2,14 +2,13 @@
 """
 Text-to-Speech Tool Module
 
-Supports seven TTS providers:
+Supports six TTS providers:
 - Edge TTS (default, free, no API key): Microsoft Edge neural voices
 - ElevenLabs (premium): High-quality voices, needs ELEVENLABS_API_KEY
 - OpenAI TTS: Good quality, needs OPENAI_API_KEY
 - MiniMax TTS: High-quality with voice cloning, needs MINIMAX_API_KEY
 - Mistral (Voxtral TTS): Multilingual, native Opus, needs MISTRAL_API_KEY
 - NeuTTS (local, free, no API key): On-device TTS via neutts_cli, needs neutts installed
-- KittenTTS (local, free, no API key): Lightweight on-device ONNX TTS via kittentts
 
 Output formats:
 - Opus (.ogg) for Telegram voice bubbles (requires ffmpeg for Edge TTS)
@@ -78,12 +77,6 @@ def _import_sounddevice():
     return sd
 
 
-def _import_kittentts():
-    """Lazy import KittenTTS. Returns the class or raises ImportError."""
-    from kittentts import KittenTTS
-    return KittenTTS
-
-
 # ===========================================================================
 # Defaults
 # ===========================================================================
@@ -93,8 +86,6 @@ DEFAULT_ELEVENLABS_VOICE_ID = "pNInz6obpgDQGcFmaJgB"  # Adam
 DEFAULT_ELEVENLABS_MODEL_ID = "eleven_multilingual_v2"
 DEFAULT_ELEVENLABS_STREAMING_MODEL_ID = "eleven_flash_v2_5"
 DEFAULT_OPENAI_MODEL = "gpt-4o-mini-tts"
-DEFAULT_KITTENTTS_MODEL = "KittenML/kitten-tts-nano-0.8-int8"  # 25MB
-DEFAULT_KITTENTTS_VOICE = "Jasper"
 DEFAULT_OPENAI_VOICE = "alloy"
 DEFAULT_OPENAI_BASE_URL = "https://api.openai.com/v1"
 DEFAULT_MINIMAX_MODEL = "speech-2.8-hd"
@@ -457,15 +448,6 @@ def _check_neutts_available() -> bool:
         return False
 
 
-def _check_kittentts_available() -> bool:
-    """Check if the kittentts engine is importable (installed locally)."""
-    try:
-        import importlib.util
-        return importlib.util.find_spec("kittentts") is not None
-    except Exception:
-        return False
-
-
 def _default_neutts_ref_audio() -> str:
     """Return path to the bundled default voice reference audio."""
     return str(Path(__file__).parent / "neutts_samples" / "jo.wav")
@@ -529,51 +511,6 @@ def _generate_neutts(text: str, output_path: str, tts_config: Dict[str, Any]) ->
     return output_path
 
 
-# ===========================================================================
-# Provider: KittenTTS (local, lightweight)
-# ===========================================================================
-
-# Module-level cache for KittenTTS model instances
-_kittentts_model_cache: Dict[str, Any] = {}
-
-
-def _generate_kittentts(text: str, output_path: str, tts_config: Dict[str, Any]) -> str:
-    """Generate speech using the local KittenTTS ONNX model."""
-    KittenTTS = _import_kittentts()
-    kt_config = tts_config.get("kittentts", {})
-    model_name = kt_config.get("model", DEFAULT_KITTENTTS_MODEL)
-    voice = kt_config.get("voice", DEFAULT_KITTENTTS_VOICE)
-    speed = kt_config.get("speed", 1.0)
-    clean_text = kt_config.get("clean_text", True)
-
-    global _kittentts_model_cache
-    if model_name not in _kittentts_model_cache:
-        logger.info("[KittenTTS] Loading model: %s", model_name)
-        _kittentts_model_cache[model_name] = KittenTTS(model_name)
-
-    model = _kittentts_model_cache[model_name]
-    audio = model.generate(text, voice=voice, speed=speed, clean_text=clean_text)
-
-    import soundfile as sf
-
-    wav_path = output_path
-    if not output_path.endswith(".wav"):
-        wav_path = output_path.rsplit(".", 1)[0] + ".wav"
-
-    sf.write(wav_path, audio, 24000)
-
-    if wav_path != output_path:
-        ffmpeg = shutil.which("ffmpeg")
-        if ffmpeg:
-            conv_cmd = [ffmpeg, "-i", wav_path, "-y", "-loglevel", "error", output_path]
-            subprocess.run(conv_cmd, check=True, timeout=30)
-            os.remove(wav_path)
-        else:
-            os.rename(wav_path, output_path)
-
-    return output_path
-
-
 # ===========================================================================
 # Main tool function
 # ===========================================================================
@@ -685,19 +622,6 @@ def text_to_speech_tool(
             logger.info("Generating speech with NeuTTS (local)...")
             _generate_neutts(text, file_str, tts_config)
 
-        elif provider == "kittentts":
-            try:
-                _import_kittentts()
-            except ImportError:
-                return json.dumps({
-                    "success": False,
-                    "error": "KittenTTS provider selected but 'kittentts' package not installed. "
-                             "Run 'hermes setup tts' and choose KittenTTS, or install manually: "
-                             "pip install https://github.com/KittenML/KittenTTS/releases/download/0.8.1/kittentts-0.8.1-py3-none-any.whl"
-                }, ensure_ascii=False)
-            logger.info("Generating speech with KittenTTS (local, lightweight)...")
-            _generate_kittentts(text, file_str, tts_config)
-
         else:
             # Default: Edge TTS (free), with NeuTTS as local fallback
             edge_available = True
@@ -734,10 +658,10 @@ def text_to_speech_tool(
                 "error": f"TTS generation produced no output (provider: {provider})"
             }, ensure_ascii=False)
 
-        # Try Opus conversion for Telegram compatibility only.
-        # Outside Telegram, preserve the caller's explicit output format.
+        # Try Opus conversion for Telegram compatibility
+        # Edge TTS outputs MP3, NeuTTS outputs WAV — both need ffmpeg conversion
         voice_compatible = False
-        if want_opus and provider in ("edge", "neutts", "minimax", "kittentts") and not file_str.endswith(".ogg"):
+        if provider in ("edge", "neutts", "minimax") and not file_str.endswith(".ogg"):
             opus_path = _convert_to_opus(file_str)
             if opus_path:
                 file_str = opus_path
@@ -818,8 +742,6 @@ def check_tts_requirements() -> bool:
         pass
     if _check_neutts_available():
         return True
-    if _check_kittentts_available():
-        return True
     return False
 
 

website/docs/user-guide/features/tts.md

@@ -10,7 +10,7 @@ Hermes Agent supports both text-to-speech output and voice message transcription
 
 ## Text-to-Speech
 
-Convert text to speech with seven providers:
+Convert text to speech with six providers:
 
 | Provider | Quality | Cost | API Key |
 |----------|---------|------|---------|
@@ -20,7 +20,6 @@ Convert text to speech with seven providers:
 | **MiniMax TTS** | Excellent | Paid | `MINIMAX_API_KEY` |
 | **Mistral (Voxtral TTS)** | Excellent | Paid | `MISTRAL_API_KEY` |
 | **NeuTTS** | Good | Free | None needed |
-| **KittenTTS** | Good | Free (local) | None needed |
 
 ### Platform Delivery
 
@@ -36,7 +35,7 @@ Convert text to speech with seven providers:
 ```yaml
 # In ~/.hermes/config.yaml
 tts:
-  provider: "edge"              # "edge" | "elevenlabs" | "openai" | "minimax" | "mistral" | "neutts" | "kittentts"
+  provider: "edge"              # "edge" | "elevenlabs" | "openai" | "minimax" | "mistral" | "neutts"
   speed: 1.0                    # Global speed multiplier (provider-specific settings override this)
   edge:
     voice: "en-US-AriaNeural"   # 322 voices, 74 languages
@@ -63,11 +62,6 @@ tts:
     ref_text: ''
     model: neuphonic/neutts-air-q4-gguf
     device: cpu
-  kittentts:
-    model: KittenML/kitten-tts-nano-0.8-int8   # 25MB int8 default; also micro and mini variants
-    voice: Jasper                               # Jasper, Bella, Luna, Bruno, Rosie, Hugo, Kiki, Leo
-    speed: 1.0
-    clean_text: true
 ```
 
 **Speed control**: The global `tts.speed` value applies to all providers by default. Each provider can override it with its own `speed` setting (e.g., `tts.openai.speed: 1.5`). Provider-specific speed takes precedence over the global value. Default is `1.0` (normal speed).
@@ -80,7 +74,6 @@ Telegram voice bubbles require Opus/OGG audio format:
 - **Edge TTS** (default) outputs MP3 and needs **ffmpeg** to convert:
 - **MiniMax TTS** outputs MP3 and needs **ffmpeg** to convert for Telegram voice bubbles
 - **NeuTTS** outputs WAV and also needs **ffmpeg** to convert for Telegram voice bubbles
-- **KittenTTS** outputs WAV and also needs **ffmpeg** to convert for Telegram voice bubbles
 
 ```bash
 # Ubuntu/Debian
@@ -93,7 +86,7 @@ brew install ffmpeg
 sudo dnf install ffmpeg
 ```
 
-Without ffmpeg, Edge TTS, MiniMax TTS, NeuTTS, and KittenTTS audio are sent as regular audio files (playable, but shown as a rectangular player instead of a voice bubble).
+Without ffmpeg, Edge TTS, MiniMax TTS, and NeuTTS audio are sent as regular audio files (playable, but shown as a rectangular player instead of a voice bubble).
 
 :::tip
 If you want voice bubbles without installing ffmpeg, switch to the OpenAI, ElevenLabs, or Mistral provider.