feat(tools): add MuAPI image and video generation tools

lib/crewai-tools/src/crewai_tools/tools/__init__.py

@@ -100,6 +100,7 @@
 from crewai_tools.tools.merge_agent_handler_tool.merge_agent_handler_tool import (
     MergeAgentHandlerTool,
 )
+from crewai_tools.tools.muapi_tool.muapi_tool import MuApiImageTool, MuApiVideoTool
 from crewai_tools.tools.mongodb_vector_search_tool import (
     MongoDBToolSchema,
     MongoDBVectorSearchConfig,
@@ -262,6 +263,8 @@
     "MongoDBToolSchema",
     "MongoDBVectorSearchConfig",
     "MongoDBVectorSearchTool",
+    "MuApiImageTool",
+    "MuApiVideoTool",
     "MultiOnTool",
     "MySQLSearchTool",
     "NL2SQLTool",

lib/crewai-tools/src/crewai_tools/tools/muapi_tool/README.MD

@@ -0,0 +1,48 @@
+# MuAPI Tools
+
+`MuApiImageTool` and `MuApiVideoTool` — generate images and videos using [muapi.ai](https://muapi.ai), a unified API with 400+ generative media models.
+
+## Setup
+
+```bash
+export MUAPI_API_KEY=your_key_here
+```
+
+Get an API key at [muapi.ai/dashboard/api-keys](https://muapi.ai/dashboard/api-keys).
+
+## Usage
+
+```python
+from crewai import Agent, Task, Crew
+from crewai_tools import MuApiImageTool, MuApiVideoTool
+
+image_tool = MuApiImageTool()
+video_tool = MuApiVideoTool()
+
+designer = Agent(
+    role="Creative Director",
+    goal="Create visual assets for marketing campaigns",
+    tools=[image_tool, video_tool],
+    backstory="You are a creative director who generates images and videos.",
+)
+
+task = Task(
+    description="Generate a photorealistic product photo of a coffee mug on a marble table.",
+    expected_output="URL of the generated image",
+    agent=designer,
+)
+
+crew = Crew(agents=[designer], tasks=[task])
+result = crew.kickoff()
+```
+
+## Supported Image Models
+
+`flux-schnell`, `flux-dev`, `flux-kontext-dev/pro/max`, `hidream-fast/dev/full`,
+`midjourney`, `gpt4o`, `gpt-image-2`, `imagen4`, `imagen4-fast`, `seedream`,
+`reve`, `ideogram`, `hunyuan`, `wan2.1`, `qwen`
+
+## Supported Video Models
+
+`veo3`, `veo3-fast`, `kling-master`, `wan2.1`, `wan2.2`, `seedance-pro`,
+`seedance-pro-fast`, `runway`, `pixverse`, `sora`, `minimax-hailuo-02-pro`

lib/crewai-tools/src/crewai_tools/tools/muapi_tool/__init__.py

@@ -0,0 +1 @@
+

lib/crewai-tools/src/crewai_tools/tools/muapi_tool/muapi_tool.py

@@ -0,0 +1,207 @@
+import json
+import time
+import urllib.error
+import urllib.request
+from typing import Any, List, Literal, Optional
+
+from crewai.tools import BaseTool, EnvVar
+from pydantic import BaseModel, Field
+
+
+BASE_URL = "https://api.muapi.ai/api/v1"
+
+IMAGE_MODELS = [
+    "flux-schnell",
+    "flux-dev",
+    "flux-kontext-dev",
+    "flux-kontext-pro",
+    "flux-kontext-max",
+    "hidream-fast",
+    "hidream-dev",
+    "hidream-full",
+    "midjourney",
+    "gpt4o",
+    "gpt-image-2",
+    "imagen4",
+    "imagen4-fast",
+    "seedream",
+    "reve",
+    "ideogram",
+    "hunyuan",
+    "wan2.1",
+    "qwen",
+]
+
+VIDEO_MODELS = [
+    "veo3",
+    "veo3-fast",
+    "kling-master",
+    "wan2.1",
+    "wan2.2",
+    "seedance-pro",
+    "seedance-pro-fast",
+    "runway",
+    "pixverse",
+    "sora",
+    "minimax-hailuo-02-pro",
+]
+
+
+class MuApiImageSchema(BaseModel):
+    """Input for MuAPI Image Generation Tool."""
+
+    prompt: str = Field(description="Text description of the image to generate.")
+    model: str = Field(
+        default="flux-schnell",
+        description=(
+            "Model to use. Options: " + ", ".join(IMAGE_MODELS)
+        ),
+    )
+    width: Optional[int] = Field(default=None, description="Image width in pixels.")
+    height: Optional[int] = Field(default=None, description="Image height in pixels.")
+
+
+class MuApiVideoSchema(BaseModel):
+    """Input for MuAPI Video Generation Tool."""
+
+    prompt: str = Field(description="Text description of the video to generate.")
+    model: str = Field(
+        default="veo3-fast",
+        description=(
+            "Model to use. Options: " + ", ".join(VIDEO_MODELS)
+        ),
+    )
+    duration: int = Field(default=5, description="Duration in seconds (3–60).")
+    aspect_ratio: Literal["16:9", "9:16", "1:1", "4:3"] = Field(
+        default="16:9", description="Aspect ratio."
+    )
+
+
+def _submit_and_poll(api_key: str, endpoint: str, payload: dict, timeout: int = 300) -> str:
+    """Submit a muapi.ai job and poll until completion, returning the output URL."""
+    headers = {"x-api-key": api_key, "Content-Type": "application/json"}
+    body = json.dumps(payload).encode()
+
+    # Submit
+    try:
+        req = urllib.request.Request(f"{BASE_URL}/{endpoint}", data=body, headers=headers, method="POST")
+        with urllib.request.urlopen(req, timeout=30) as resp:
+            data = json.loads(resp.read())
+    except urllib.error.HTTPError as exc:
+        raise RuntimeError(f"MuAPI submit failed [{exc.code}]: {exc.read().decode(errors='replace')}") from exc
+    except urllib.error.URLError as exc:
+        raise RuntimeError(f"MuAPI submit connection error: {exc.reason}") from exc
+
+    request_id = data.get("request_id")
+    if not request_id:
+        raise RuntimeError(f"MuAPI did not return a request_id: {data}")
+
+    # Poll
+    deadline = time.time() + timeout
+    while time.time() < deadline:
+        time.sleep(3)
+        try:
+            poll_req = urllib.request.Request(
+                f"{BASE_URL}/predictions/{request_id}/result",
+                headers={"x-api-key": api_key},
+            )
+            with urllib.request.urlopen(poll_req, timeout=15) as resp:
+                result = json.loads(resp.read())
+        except urllib.error.HTTPError as exc:
+            raise RuntimeError(f"MuAPI poll failed [{exc.code}]: {exc.read().decode(errors='replace')}") from exc
+        except urllib.error.URLError as exc:
+            raise RuntimeError(f"MuAPI poll connection error: {exc.reason}") from exc
+
+        status = result.get("status", "pending")
+        if status == "completed":
+            outputs = result.get("outputs", [])
+            if not outputs:
+                raise RuntimeError("Generation completed but returned no outputs")
+            return outputs[0]
+        if status in ("failed", "cancelled"):
+            raise RuntimeError(f"Generation {status}: {result.get('error', '')}")
+
+    raise TimeoutError(f"Generation timed out after {timeout}s")
+
+
+class MuApiImageTool(BaseTool):
+    """Generate images using muapi.ai — a unified API for 400+ image generation models."""
+
+    name: str = "MuAPI Image Generator"
+    description: str = (
+        "Generates images from text prompts using muapi.ai. "
+        "Supports Flux, Midjourney, GPT-4o Image, Google Imagen 4, "
+        "Seedream, HiDream, Reve, Ideogram, and more. "
+        "Returns the URL of the generated image."
+    )
+    args_schema: type[BaseModel] = MuApiImageSchema
+
+    env_vars: List[EnvVar] = [
+        EnvVar(
+            name="MUAPI_API_KEY",
+            description="API key for muapi.ai. Get one at https://muapi.ai/dashboard/api-keys",
+            required=True,
+        )
+    ]
+
+    def _run(self, **kwargs: Any) -> str:
+        import os
+
+        api_key = os.environ.get("MUAPI_API_KEY", "")
+        if not api_key:
+            return "Error: MUAPI_API_KEY environment variable not set."
+
+        prompt = kwargs.get("prompt", "")
+        model = kwargs.get("model", "flux-schnell")
+        payload: dict = {"prompt": prompt}
+        if kwargs.get("width"):
+            payload["width"] = kwargs["width"]
+        if kwargs.get("height"):
+            payload["height"] = kwargs["height"]
+
+        try:
+            url = _submit_and_poll(api_key, model, payload)
+            return json.dumps({"image_url": url, "model": model, "prompt": prompt})
+        except Exception as e:
+            return f"Error generating image: {e}"
+
+
+class MuApiVideoTool(BaseTool):
+    """Generate videos using muapi.ai — a unified API for 400+ video generation models."""
+
+    name: str = "MuAPI Video Generator"
+    description: str = (
+        "Generates short videos from text prompts using muapi.ai. "
+        "Supports Veo3, Kling, Wan, Seedance, Runway, Pixverse, Sora, and more. "
+        "Returns the URL of the generated MP4 video."
+    )
+    args_schema: type[BaseModel] = MuApiVideoSchema
+
+    env_vars: List[EnvVar] = [
+        EnvVar(
+            name="MUAPI_API_KEY",
+            description="API key for muapi.ai. Get one at https://muapi.ai/dashboard/api-keys",
+            required=True,
+        )
+    ]
+
+    def _run(self, **kwargs: Any) -> str:
+        import os
+
+        api_key = os.environ.get("MUAPI_API_KEY", "")
+        if not api_key:
+            return "Error: MUAPI_API_KEY environment variable not set."
+
+        prompt = kwargs.get("prompt", "")
+        model = kwargs.get("model", "veo3-fast")
+        payload: dict = {
+            "prompt": prompt,
+            "duration": kwargs.get("duration", 5),
+            "aspect_ratio": kwargs.get("aspect_ratio", "16:9"),
+        }
+
+        try:
+            url = _submit_and_poll(api_key, model, payload, timeout=600)
+            return json.dumps({"video_url": url, "model": model, "prompt": prompt})
+        except Exception as e:
+            return f"Error generating video: {e}"