lcode-cli 0.20.0__tar.gz → 0.22.0__tar.gz

{lcode_cli-0.20.0 → lcode_cli-0.22.0}/CHANGELOG.md

@@ -6,6 +6,30 @@ All notable changes to lcode are documented here. The format follows
 
 ## [Unreleased]
 
+## [0.22.0] - 2026-10-06
+
+### Added
+
+- `lcode mcp add xai`: images, image edits, videos and speech with xAI's Grok Imagine and text to
+  speech APIs, in the cloud and billed to your xAI account. It asks for your xAI API key, and runs
+  `lcode-mcp-xai`, a new MCP server that ships with lcode, with the tools `generate_image`,
+  `edit_image`, `generate_video` (waits for the video and downloads it), `video_status` and
+  `text_to_speech`. Files are saved in `generated/` in the project or where you ask. See
+  [Images, videos and speech with xAI](https://nasser1941.github.io/lcode/mcp/#images-videos-and-speech-with-xai).
+
+## [0.21.0] - 2026-10-06
+
+### Added
+
+- `lcode update` updates lcode the way it was installed (uv tool, pipx, Homebrew, pip, or
+  `git pull` for a checkout), and Ollama when it runs in Docker. The newest Ollama release comes
+  from Docker Hub, not a local `latest` tag, and is pulled by its version tag (such as
+  `ollama/ollama:0.35.1`) so `docker ps` shows the version. A container from Docker Compose gets
+  the new tag in its compose file and is recreated with `docker compose up`; any other container is
+  recreated with the same settings, and the old one comes back if the new one doesn't start.
+  `lcode update --check` only shows which versions are out. See
+  [Updating](https://nasser1941.github.io/lcode/installation/#updating).
+
 ## [0.20.0] - 2026-10-06
 
 ### Added
@@ -467,7 +491,9 @@ First public release.
 - `AGENTS.md` project instructions and `/init` to generate them.
 - One-line installer for Linux and macOS.
 
-[Unreleased]: https://github.com/nasser1941/lcode/compare/v0.20.0...HEAD
+[Unreleased]: https://github.com/nasser1941/lcode/compare/v0.22.0...HEAD
+[0.22.0]: https://github.com/nasser1941/lcode/compare/v0.21.0...v0.22.0
+[0.21.0]: https://github.com/nasser1941/lcode/compare/v0.20.0...v0.21.0
 [0.20.0]: https://github.com/nasser1941/lcode/compare/v0.19.0...v0.20.0
 [0.19.0]: https://github.com/nasser1941/lcode/compare/v0.18.0...v0.19.0
 [0.18.0]: https://github.com/nasser1941/lcode/compare/v0.17.0...v0.18.0

{lcode_cli-0.20.0 → lcode_cli-0.22.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: lcode-cli
-Version: 0.20.0
+Version: 0.22.0
 Summary: A local-first terminal coding agent powered by open-weight models on your own GPU or Mac, via Ollama.
 Project-URL: Homepage, https://nasser1941.github.io/lcode/
 Project-URL: Documentation, https://nasser1941.github.io/lcode/
@@ -116,7 +116,8 @@ through [Ollama](https://ollama.com), and when it needs current information it c
   (`lcode mcp add atlassian`), image generation with ComfyUI, or any other MCP server. Browser sign-in (OAuth) is built in.
 - **Sees images.** Attach a screenshot or mockup with `@path` and lcode looks at it, using the
   vision part of your model. With ComfyUI (`lcode mcp add comfyui`) it can also generate and edit
-  images with local models such as FLUX, SDXL and Qwen-Image.
+  images with local models such as FLUX, SDXL and Qwen-Image, or, in the cloud, images, videos
+  and speech with xAI's Grok Imagine (`lcode mcp add xai`).
 - **Web search when needed.** Looks up the latest versions, docs and error messages with Ollama web
   search, Brave, Tavily or your own SearXNG, and reads pages as clean text.
 - **Private.** The model runs on your machine, lcode never uploads your files and has no telemetry.

{lcode_cli-0.20.0 → lcode_cli-0.22.0}/README.md

@@ -80,7 +80,8 @@ through [Ollama](https://ollama.com), and when it needs current information it c
   (`lcode mcp add atlassian`), image generation with ComfyUI, or any other MCP server. Browser sign-in (OAuth) is built in.
 - **Sees images.** Attach a screenshot or mockup with `@path` and lcode looks at it, using the
   vision part of your model. With ComfyUI (`lcode mcp add comfyui`) it can also generate and edit
-  images with local models such as FLUX, SDXL and Qwen-Image.
+  images with local models such as FLUX, SDXL and Qwen-Image, or, in the cloud, images, videos
+  and speech with xAI's Grok Imagine (`lcode mcp add xai`).
 - **Web search when needed.** Looks up the latest versions, docs and error messages with Ollama web
   search, Brave, Tavily or your own SearXNG, and reads pages as clean text.
 - **Private.** The model runs on your machine, lcode never uploads your files and has no telemetry.

{lcode_cli-0.20.0 → lcode_cli-0.22.0}/docs/installation.md

@@ -161,11 +161,36 @@ lmstudio` (or `llama.cpp`, `vllm`, `mlx`). See [Other model servers](servers.md)
 Use [WSL2](https://learn.microsoft.com/windows/wsl/install) with Ubuntu and follow the Ubuntu
 steps. NVIDIA GPUs work inside WSL2 with the regular Windows driver.
 
-## Upgrade and uninstall
+## Updating
 
 ```bash
-uv tool upgrade lcode-cli        # upgrade (pipx: pipx upgrade lcode-cli; Homebrew: brew upgrade lcode)
-uv tool uninstall lcode-cli      # remove lcode (Homebrew: brew uninstall lcode)
+lcode update            # update lcode, and Ollama if it runs in Docker
+lcode update --check    # only show which versions are out
+```
+
+- **lcode** updates the way you installed it: `uv tool upgrade`, `pipx upgrade`, `brew upgrade`
+  or `pip`. If it runs from a git checkout, `git pull` (on `main`, without local changes).
+- **Ollama in a Docker container** is updated to the newest release on Docker Hub. lcode looks
+  that up instead of trusting a local `latest` tag, which can be months old, and skips release
+  candidates. It pulls the version's own tag, such as `ollama/ollama:0.35.1`, so `docker ps` shows
+  which version runs. A ROCm image stays ROCm.
+  - **A container from Docker Compose:** lcode changes the `image:` line in its compose file and
+    runs `docker compose up -d` for that service, so Compose keeps managing it.
+  - **Any other container** is recreated with the same name, ports, volumes, environment, GPUs,
+    network and restart policy. The old one is kept until the new one answers with the new
+    version, and is put back if it doesn't.
+  - Your models stay: they're in the container's volume.
+- **Ollama installed without Docker** gets the command that updates it: on Linux,
+  `curl -fsSL https://ollama.com/install.sh | sh`; on macOS, `brew upgrade ollama` or the app
+  itself.
+
+lcode shows what it will do and asks first. `lcode update lcode` or `lcode update ollama` updates
+only one; `-y` doesn't ask.
+
+## Uninstall
+
+```bash
+uv tool uninstall lcode-cli      # remove lcode (pipx: pipx uninstall lcode-cli; Homebrew: brew uninstall lcode)
 rm -rf ~/.config/lcode ~/.local/state/lcode    # remove settings and saved sessions
 ollama rm lcode-qwen3.6-35b qwen3.6:35b-a3b-coding   # remove downloaded models
 ```

{lcode_cli-0.20.0 → lcode_cli-0.22.0}/docs/mcp.md

@@ -29,6 +29,7 @@ and find the code for the top one."*
 | `github` | Repositories, issues, pull requests, Actions | a GitHub token (or the GitHub CLI) |
 | `playwright` | A real (headless) browser: open pages, click, fill forms | `npx` |
 | `comfyui` | Generate and edit images with local models (FLUX, SDXL, SD 1.5, Qwen-Image) | `uvx` and [ComfyUI](#images-with-comfyui) running locally |
+| `xai` | Generate and edit images, and generate videos and speech, with xAI's Grok Imagine (paid, in the cloud) | `uvx`, an xAI API key ([below](#images-videos-and-speech-with-xai)) |
 | `context7` | Up-to-date docs and examples for thousands of libraries | nothing (an API key is optional) |
 | `sentry` | Errors, issues, traces and releases | a Sentry account (browser sign-in) |
 | `postgres` | Schemas, read-only queries, query performance | `uvx`, a connection URL |
@@ -137,6 +138,36 @@ size, prompt), which can take several minutes of trial and error. Once an image
 lcode to save that workflow in your project (for example `assets/icon.workflow.json`) and reuse it:
 later images are a single `run_workflow` call.
 
+### Images, videos and speech with xAI
+
+xAI's [Grok Imagine](https://docs.x.ai/developers/model-capabilities/imagine) and
+[text to speech](https://docs.x.ai/developers/models/text-to-speech) APIs make images, videos and
+speech in the cloud: no GPU needed, but each call is billed to your xAI account. lcode ships an MCP
+server for them, `lcode-mcp-xai`.
+
+```bash
+lcode mcp add xai             # asks for your xAI API key (create one at https://console.x.ai)
+```
+
+The key is typed hidden and kept in `~/.config/lcode/mcp.json`, readable only by you. If
+`XAI_API_KEY` is already set in your environment, lcode uses that and doesn't store the key.
+
+| Tool | | Costs |
+|---|---|---|
+| `generate_image` | Images from a description: 1–10 at a time, aspect ratio, 1k or 2k | $0.02–0.05 an image |
+| `edit_image` | Change a local image as described | the same |
+| `generate_video` | A 1–15 second MP4 from a description, optionally starting from a local image, up to 1080p. Waits for it (a minute or more) | $0.02–0.08 a second |
+| `video_status` | Fetch a video that wasn't ready within 10 minutes | |
+| `text_to_speech` | Speech in one of 26 voices and 20 languages, as MP3 or WAV | per character |
+
+Files are saved in `generated/` in the project, or where you ask (*"generate an app icon and save
+it as assets/icon.png"*). xAI has no music generation API.
+
+Your prompts, and the images you edit or animate, go to xAI. lcode asks before each call, since each
+one costs money. To stop asking for some tools, list them under `"allow"` for the server in
+`mcp.json`, for example `"allow": ["generate_image", "text_to_speech"]`. xAI applies its own usage
+policies to what it generates; when it refuses a prompt, lcode passes its message to the model.
+
 ### Metabase
 
 Metabase has its own MCP server inside every instance (Metabase 60 and later), at

{lcode_cli-0.20.0 → lcode_cli-0.22.0}/docs/quickstart.md

@@ -12,7 +12,7 @@ lcode
 
 ```text
 ╭──────────────────────────────────────────────╮
-│ lcode v0.20.0 — local coding agent           │
+│ lcode v0.22.0 — local coding agent           │
 │                                              │
 │ model    lcode-qwen3.6-35b                   │
 │ context  256K tokens                         │

{lcode_cli-0.20.0 → lcode_cli-0.22.0}/docs/usage.md

@@ -11,6 +11,7 @@ lcode config [set|unset]     # show or change settings
 lcode bench [models]         # score models on small coding tasks on this machine
 lcode mcp [catalog|add|list] # connect MCP servers: Jira, GitHub, AWS, databases… (see MCP servers)
 lcode acp                    # run inside an editor over the Agent Client Protocol (see Editors)
+lcode update [--check]       # update lcode, and Ollama when it runs in Docker (see Installation)
 ```
 
 | Option | Meaning |

{lcode_cli-0.20.0 → lcode_cli-0.22.0}/pyproject.toml

@@ -49,6 +49,7 @@ Changelog = "https://github.com/nasser1941/lcode/blob/main/CHANGELOG.md"
 lcode = "lcode.cli:main"
 lcode-mcp-valohai = "lcode.servers.valohai:main"
 lcode-mcp-encord = "lcode.servers.encord:main"
+lcode-mcp-xai = "lcode.servers.xai:main"
 
 [dependency-groups]
 dev = ["pytest>=8", "ruff>=0.6"]
@@ -87,6 +88,8 @@ ignore = [
 "src/lcode/gitflow.py" = ["E501"]
 "src/lcode/action.py" = ["E501"]
 "src/lcode/acp.py" = ["E501"]
+"src/lcode/update.py" = ["E501"]
+"src/lcode/servers/xai.py" = ["E501"]
 
 [tool.pytest.ini_options]
 testpaths = ["tests"]

{lcode_cli-0.20.0 → lcode_cli-0.22.0}/src/lcode/__init__.py

@@ -3,7 +3,7 @@
 From a program: `from lcode import Session` (see lcode.api).
 """
 
-__version__ = "0.20.0"
+__version__ = "0.22.0"
 
 API = ("Session", "Result", "Options", "SetupError")
 

{lcode_cli-0.20.0 → lcode_cli-0.22.0}/src/lcode/cli.py

@@ -678,7 +678,7 @@ def build_parser() -> argparse.ArgumentParser:
     parser = argparse.ArgumentParser(
         prog="lcode",
         description="A local-first terminal coding agent powered by open-weight models via Ollama.",
-        epilog="Subcommands: lcode setup | models | doctor | config | bench | mcp | index "
+        epilog="Subcommands: lcode setup | models | doctor | config | bench | mcp | index | update "
         "(lcode <subcommand> --help). "
         "Docs: https://nasser1941.github.io/lcode/",
     )
@@ -774,6 +774,15 @@ def build_subparsers() -> dict[str, argparse.ArgumentParser]:
     from lcode.mcp.commands import build_parser as mcp_parser
 
     subs["mcp"] = mcp_parser()
+    p = argparse.ArgumentParser(
+        prog="lcode update",
+        description="Update lcode, and Ollama when it runs in Docker (to the newest version on Docker Hub, pinned "
+        "to its version tag).",
+    )
+    p.add_argument("what", nargs="?", choices=["lcode", "ollama"], help="only this one (default: both)")
+    p.add_argument("--check", action="store_true", help="only show which versions are out")
+    p.add_argument("-y", "--yes", action="store_true", help="don't ask before updating")
+    subs["update"] = p
     subs["acp"] = argparse.ArgumentParser(
         prog="lcode acp",
         description="Run as the agent inside an editor, over the Agent Client Protocol on stdin and stdout "
@@ -787,6 +796,12 @@ def build_subparsers() -> dict[str, argparse.ArgumentParser]:
     return subs
 
 
+def cmd_update(args) -> None:
+    from lcode import update
+
+    sys.exit(update.command(console, args.what or "", check=args.check, yes=args.yes))
+
+
 def cmd_acp(args) -> None:
     from lcode import acp
 
@@ -863,6 +878,7 @@ COMMANDS = {
     "index": cmd_index,
     "action": cmd_action,
     "acp": cmd_acp,
+    "update": cmd_update,
 }
 
 

{lcode_cli-0.20.0 → lcode_cli-0.22.0}/src/lcode/mcp/catalog.toml

@@ -281,3 +281,24 @@ secret = true
 var = "VALOHAI_HOST"
 prompt = "Valohai address (Enter for app.valohai.com; for a self-hosted installation, its URL)"
 optional = true
+
+[xai]
+name = "xAI Grok Imagine"
+description = "Generate and edit images, and generate videos and speech, with xAI's Grok Imagine and voice API (paid, in the cloud)"
+homepage = "https://nasser1941.github.io/lcode/mcp/#xai"
+requires = ["uvx"]
+setup = """Uses xAI's API: your prompts, and images you edit or animate, go to xAI, and every call is billed to your xAI
+account (an image $0.02-0.05, video $0.02-0.08 a second, speech per character). Create an API key at https://console.x.ai.
+Files are saved in generated/ in the project. lcode asks before each call; to stop asking for some tools, list them
+under "allow" for this server in mcp.json. xAI has no music generation."""
+
+[xai.server]
+command = "uvx"
+args = ["--isolated", "--python", "3.12", "--from", "lcode-cli", "lcode-mcp-xai"]
+env = { XAI_API_KEY = "${XAI_API_KEY}" }
+timeout = 900  # a video takes minutes
+
+[[xai.inputs]]
+var = "XAI_API_KEY"
+prompt = "xAI API key (create one at https://console.x.ai)"
+secret = true

lcode_cli-0.22.0/src/lcode/servers/xai.py

@@ -0,0 +1,289 @@
+"""An MCP server for xAI's Grok Imagine and voice APIs: images, image edits, videos and speech.
+
+xAI has an HTTP API but no MCP server, so lcode ships this one. It needs an API key (XAI_API_KEY,
+from https://console.x.ai) and saves what it makes in the project, in `generated/` unless a path is
+given. The prompts, and any image it edits or animates, go to xAI, which bills per image, per
+second of video and per character of speech, and applies its own usage policies.
+
+Run: lcode-mcp-xai   (lcode mcp add xai sets it up)
+"""
+
+from __future__ import annotations
+
+import argparse
+import base64
+import mimetypes
+import os
+import re
+import time
+from pathlib import Path
+
+import requests
+
+from lcode import __version__
+from lcode.mcp.server import Server, ToolFailure
+
+API = "https://api.x.ai/v1"
+TIMEOUT = 120
+IMAGE_MODELS = ["grok-imagine-image-2.0", "grok-imagine-image-quality", "grok-imagine-image"]
+VIDEO_MODELS = ["grok-imagine-video-1.5", "grok-imagine-video-1.5-lite", "grok-imagine-video"]
+ASPECT_RATIOS = ["auto", "1:1", "16:9", "9:16", "4:3", "3:4", "3:2", "2:3", "2:1", "1:2", "21:9"]
+VOICES = ["eve", "ara", "leo", "rex", "sal", "luna", "iris", "orion", "helix", "zagan", "carina", "altair", "zenith",
+          "perseus", "helios", "lux", "kepler", "rigel", "cosmo", "celeste", "ursa", "sirius", "lumen", "castor",
+          "naksh", "atlas"]  # fmt: skip
+AUDIO_FORMATS = {"mp3": ".mp3", "wav": ".wav"}
+VIDEO_WAIT = 600  # seconds generate_video waits before handing back the request id
+POLL = 5
+
+
+class XAI:
+    def __init__(self, key: str, base: str = API, output: Path | None = None):
+        self.base = base.rstrip("/")
+        self.output = output or Path("generated")
+        self.session = requests.Session()
+        self.session.headers.update({"Authorization": f"Bearer {key}", "Content-Type": "application/json"})
+
+    def request(self, method: str, path: str, **kwargs) -> requests.Response:
+        try:
+            r = self.session.request(method, f"{self.base}{path}", timeout=TIMEOUT, **kwargs)
+        except requests.RequestException as e:
+            raise ToolFailure(f"can't reach xAI: {e}") from e
+        if r.status_code in (401, 403):
+            raise ToolFailure(
+                f"xAI refused the request ({r.status_code}): {_error(r)}. Check XAI_API_KEY (console.x.ai)."
+            )
+        if r.status_code == 429:
+            raise ToolFailure(f"xAI's rate limit or credit limit was reached: {_error(r)}")
+        if r.status_code >= 400:
+            raise ToolFailure(f"xAI error {r.status_code}: {_error(r)}")
+        return r
+
+    def post(self, path: str, body: dict) -> dict:
+        return self.request("POST", path, json=body).json()
+
+    def download(self, url: str) -> bytes:
+        try:
+            r = requests.get(url, timeout=TIMEOUT)
+            r.raise_for_status()
+        except requests.RequestException as e:
+            raise ToolFailure(f"couldn't download the result from xAI: {e}") from e
+        return r.content
+
+    def target(self, path: str, stem: str, suffix: str) -> Path:
+        """Where to save: the given path, or a new file in the output folder."""
+        if path:
+            target = Path(path).expanduser()
+            if not target.suffix:
+                target = target.with_suffix(suffix)
+        else:
+            name = re.sub(r"[^a-z0-9]+", "-", stem.lower()).strip("-")[:40] or "xai"
+            target = self.output / f"{name}-{time.strftime('%Y%m%d-%H%M%S')}{suffix}"
+        target.parent.mkdir(parents=True, exist_ok=True)
+        n, unique = 2, target
+        while unique.exists():
+            unique = target.with_name(f"{target.stem}-{n}{target.suffix}")
+            n += 1
+        return unique
+
+
+def _error(r: requests.Response) -> str:
+    try:
+        data = r.json()
+    except ValueError:
+        return r.text[:300]
+    error = data.get("error") if isinstance(data, dict) else None
+    if isinstance(error, dict):
+        return str(error.get("message") or error)[:300]
+    return str(error or data)[:300]
+
+
+def data_url(path: str) -> str:
+    """A local image as a data: URL, for edits and image-to-video."""
+    file = Path(path).expanduser()
+    if not file.is_file():
+        raise ToolFailure(f"{path} doesn't exist")
+    mime = mimetypes.guess_type(file.name)[0] or "image/png"
+    if mime not in ("image/png", "image/jpeg", "image/webp"):
+        raise ToolFailure(f"{path} isn't a PNG, JPEG or WebP image")
+    return f"data:{mime};base64,{base64.b64encode(file.read_bytes()).decode()}"
+
+
+def image_suffix(data: bytes) -> str:
+    if data[:3] == b"\xff\xd8\xff":
+        return ".jpg"
+    if data[:4] == b"RIFF" and data[8:12] == b"WEBP":
+        return ".webp"
+    return ".png"
+
+
+def save_images(api: XAI, reply: dict, prompt: str, path: str) -> list[Path]:
+    saved = []
+    items = reply.get("data") or []
+    if not items:
+        raise ToolFailure(f"xAI returned no image: {str(reply)[:300]}")
+    for i, item in enumerate(items):
+        data = base64.b64decode(item["b64_json"]) if item.get("b64_json") else api.download(item["url"])
+        many = path and len(items) > 1
+        target = api.target(f"{Path(path).with_suffix('')}-{i + 1}" if many else path, prompt, image_suffix(data))
+        target.write_bytes(data)
+        saved.append(target)
+    return saved
+
+
+def build(api: XAI) -> Server:
+    server = Server(
+        "xai",
+        __version__,
+        "xAI's Grok Imagine and voice: generate and edit images, generate videos (from text or a start image) "
+        "and speech. Results are saved as files in the project; each call costs money on the user's xAI account. "
+        "xAI has no music generation.",
+    )
+    path_arg = {"type": "string", "description": "Where to save it (default: a new file in generated/)"}
+
+    @server.tool(
+        "generate_image",
+        "Generate images from a description with Grok Imagine and save them. Costs $0.02–0.05 per image.",
+        {
+            "prompt": {"type": "string", "description": "What the image shows, in detail"},
+            "n": {"type": "integer", "description": "How many images, 1–10 (default 1)"},
+            "aspect_ratio": {"type": "string", "enum": ASPECT_RATIOS, "description": "Default auto"},
+            "resolution": {"type": "string", "enum": ["1k", "2k"], "description": "Default 1k"},
+            "model": {"type": "string", "enum": IMAGE_MODELS, "description": f"Default {IMAGE_MODELS[0]}"},
+            "path": path_arg,
+        },
+        ["prompt"],
+    )
+    def generate_image(prompt: str, n: int = 1, aspect_ratio: str = "auto", resolution: str = "1k",
+                       model: str = IMAGE_MODELS[0], path: str = "") -> str:  # fmt: skip
+        body = {"model": model, "prompt": prompt, "n": max(1, min(10, int(n))), "response_format": "b64_json"}
+        if aspect_ratio and aspect_ratio != "auto":
+            body["aspect_ratio"] = aspect_ratio
+        if resolution and resolution != "1k":
+            body["resolution"] = resolution
+        saved = save_images(api, api.post("/images/generations", body), prompt, path)
+        return f"Saved {len(saved)} image(s) from {model}:\n" + "\n".join(f"- {p}" for p in saved)
+
+    @server.tool(
+        "edit_image",
+        "Change an existing image (a local file) as described, with Grok Imagine, and save the result.",
+        {
+            "image": {"type": "string", "description": "Path of the image to edit (PNG, JPEG or WebP)"},
+            "prompt": {"type": "string", "description": "What to change"},
+            "model": {"type": "string", "enum": IMAGE_MODELS, "description": f"Default {IMAGE_MODELS[0]}"},
+            "path": path_arg,
+        },
+        ["image", "prompt"],
+    )
+    def edit_image(image: str, prompt: str, model: str = IMAGE_MODELS[0], path: str = "") -> str:
+        body = {"model": model, "prompt": prompt, "image": {"url": data_url(image), "type": "image_url"},
+                "response_format": "b64_json"}  # fmt: skip
+        saved = save_images(api, api.post("/images/edits", body), prompt, path)
+        return f"Saved the edited image: {saved[0]}"
+
+    def finish_video(request_id: str, prompt: str, path: str, wait: float) -> str:
+        deadline = time.monotonic() + wait
+        while True:
+            state = api.request("GET", f"/videos/{request_id}").json()
+            status = state.get("status")
+            if status == "done":
+                video = state.get("video") or {}
+                target = api.target(path, prompt or request_id, ".mp4")
+                target.write_bytes(api.download(video["url"]))
+                return f"Saved the video ({video.get('duration', '?')} s, {state.get('model', '')}): {target}"
+            if status in ("failed", "expired"):
+                raise ToolFailure(f"xAI couldn't make the video ({status}): {state.get('error') or state}")
+            if time.monotonic() > deadline:
+                return (
+                    f"The video isn't ready yet (request {request_id}). Check later with "
+                    f'video_status(request_id="{request_id}"), which downloads it when it\'s done.'
+                )
+            time.sleep(POLL)
+
+    @server.tool(
+        "generate_video",
+        "Generate a video with Grok Imagine, from a description and optionally a start image, and save it as "
+        "MP4. Takes a minute or more. Costs $0.02–0.08 per second of video.",
+        {
+            "prompt": {"type": "string", "description": "What happens in the video"},
+            "image": {"type": "string", "description": "A local image to start from (optional)"},
+            "duration": {"type": "integer", "description": "Seconds, 1–15 (default 8)"},
+            "aspect_ratio": {
+                "type": "string",
+                "enum": [r for r in ASPECT_RATIOS if r not in ("auto", "2:1", "1:2", "21:9")],
+            },
+            "resolution": {"type": "string", "enum": ["480p", "720p", "1080p"], "description": "Default 720p"},
+            "model": {"type": "string", "enum": VIDEO_MODELS, "description": f"Default {VIDEO_MODELS[0]}"},
+            "path": path_arg,
+        },
+        ["prompt"],
+    )
+    def generate_video(prompt: str, image: str = "", duration: int = 8, aspect_ratio: str = "16:9",
+                       resolution: str = "720p", model: str = VIDEO_MODELS[0], path: str = "") -> str:  # fmt: skip
+        body: dict = {"model": model, "prompt": prompt, "duration": max(1, min(15, int(duration))),
+                      "aspect_ratio": aspect_ratio, "resolution": resolution}  # fmt: skip
+        if image:
+            body["image"] = {"url": data_url(image)}
+        request_id = api.post("/videos/generations", body).get("request_id")
+        if not request_id:
+            raise ToolFailure("xAI didn't start the video")
+        return finish_video(str(request_id), prompt, path, VIDEO_WAIT)
+
+    @server.tool(
+        "video_status",
+        "Check a video that generate_video started and that wasn't ready yet; saves it when it's done.",
+        {"request_id": {"type": "string"}, "path": path_arg},
+        ["request_id"],
+    )
+    def video_status(request_id: str, path: str = "") -> str:
+        return finish_video(request_id, "", path, 0)
+
+    @server.tool(
+        "text_to_speech",
+        "Turn text into speech with an xAI voice and save it as an audio file. Billed per character.",
+        {
+            "text": {"type": "string", "description": "What to say (up to 60,000 characters)"},
+            "voice": {"type": "string", "enum": VOICES, "description": "Default eve"},
+            "language": {"type": "string", "description": "BCP-47 code such as en, de, fr, ja, or auto (default)"},
+            "format": {"type": "string", "enum": list(AUDIO_FORMATS), "description": "Default mp3"},
+            "speed": {"type": "number", "description": "1.0 is normal"},
+            "path": path_arg,
+        },
+        ["text"],
+    )
+    def text_to_speech(text: str, voice: str = "eve", language: str = "auto", format: str = "mp3",
+                       speed: float = 1.0, path: str = "") -> str:  # fmt: skip
+        codec = format if format in AUDIO_FORMATS else "mp3"
+        body = {"text": text, "voice_id": voice.lower(), "language": language or "auto",
+                "output_format": {"codec": codec}, "speed": float(speed or 1.0)}  # fmt: skip
+        r = api.request("POST", "/tts", json=body)
+        duration = ""
+        if "json" in r.headers.get("Content-Type", ""):
+            data = r.json()
+            audio = base64.b64decode(data.get("audio", ""))
+            duration = f", {data['duration']:.1f} s" if isinstance(data.get("duration"), (int, float)) else ""
+        else:
+            audio = r.content
+        if not audio:
+            raise ToolFailure("xAI returned no audio")
+        target = api.target(path, text[:40], AUDIO_FORMATS[codec])
+        target.write_bytes(audio)
+        return f"Saved the speech ({voice}{duration}): {target}"
+
+    return server
+
+
+def main(argv: list[str] | None = None) -> None:
+    parser = argparse.ArgumentParser(
+        prog="lcode-mcp-xai", description="MCP server for xAI's image, video and speech APIs (stdio)."
+    )
+    parser.add_argument("--output", default=os.environ.get("LCODE_XAI_OUTPUT", "generated"),
+                        help="folder for the files it makes (default: generated/ in the project)")  # fmt: skip
+    args = parser.parse_args(argv)
+    key = os.environ.get("XAI_API_KEY")
+    if not key:
+        parser.error("set XAI_API_KEY to an xAI API key (https://console.x.ai)")
+    build(XAI(key, os.environ.get("XAI_API_BASE", API), Path(args.output))).serve()
+
+
+if __name__ == "__main__":
+    main()