Skip to content

"CLI-Anything: Making ALL Software Agent-Native" -- CLI-Hub: https://clianything.cc/

Python

feat: add ComfyUI agent harness

Andrew Barnes08b212757812
diff --git a/comfyui/agent-harness/cli_anything/comfyui/core/images.py b/comfyui/agent-harness/cli_anything/comfyui/core/images.py
new file mode 100644
index 000000000..45d07939c
--- /dev/null
+++ b/comfyui/agent-harness/cli_anything/comfyui/core/images.py
@@ -0,0 +1,137 @@
+"""Image output management — download and list generated images.
+
+Covers:
+- Listing output images from a prompt's history
+- Downloading images from ComfyUI's /view endpoint
+- Saving images to local disk
+"""
+
+from pathlib import Path
+
+from cli_anything.comfyui.utils.comfyui_backend import api_get_raw
+from cli_anything.comfyui.core.queue import get_prompt_history
+
+
+def list_output_images(base_url: str, prompt_id: str) -> list[dict]:
+ """List all output images for a completed prompt.
+
+ Args:
+ base_url: ComfyUI server base URL.
+ prompt_id: The prompt ID returned from queue_prompt().
+
+ Returns:
+ List of image dicts with 'filename', 'subfolder', 'type', and 'node_id'.
+
+ Raises:
+ RuntimeError: If the prompt is not found or not yet completed.
+ """
+ history = get_prompt_history(base_url, prompt_id)
+
+ outputs = history.get("outputs", [])
+ if not outputs:
+ status = history.get("status", "unknown")
+ if not history.get("completed", False):
+ raise RuntimeError(
+ f"Prompt {prompt_id} has not completed yet (status: {status}). "
+ "Wait for generation to finish before listing images."
+ )
+ return []
+
+ return outputs
+
+
+def download_image(
+ base_url: str,
+ filename: str,
+ output_path: str,
+ subfolder: str = "",
+ image_type: str = "output",
+ overwrite: bool = False,
+) -> dict:
+ """Download a single output image from ComfyUI.
+
+ Args:
+ base_url: ComfyUI server base URL.
+ filename: Image filename (e.g., 'ComfyUI_00001_.png').
+ output_path: Local path to save the image.
+ subfolder: Subfolder within ComfyUI's output directory (usually empty).
+ image_type: Image type — 'output', 'input', or 'temp'.
+ overwrite: If False, raise RuntimeError if output_path already exists.
+
+ Returns:
+ Dict with 'status', 'path', and 'size_bytes'.
+
+ Raises:
+ RuntimeError: If the file exists and overwrite is False, or download fails.
+ """
+ dest = Path(output_path)
+
+ if dest.exists() and not overwrite:
+ raise RuntimeError(
+ f"Output file already exists: {output_path}. "
+ "Use overwrite=True to replace it."
+ )
+
+ params = {
+ "filename": filename,
+ "type": image_type,
+ }
+ if subfolder:
+ params["subfolder"] = subfolder
+
+ image_bytes = api_get_raw(base_url, "/view", params=params)
+
+ dest.parent.mkdir(parents=True, exist_ok=True)
+ dest.write_bytes(image_bytes)
+
+ return {
+ "status": "downloaded",
+ "path": str(dest.resolve()),
+ "filename": filename,
+ "size_bytes": len(image_bytes),
+ }
+
+
+def download_prompt_images(
+ base_url: str,
+ prompt_id: str,
+ output_dir: str,
+ overwrite: bool = False,
+) -> list[dict]:
+ """Download all output images for a completed prompt to a directory.
+
+ Args:
+ base_url: ComfyUI server base URL.
+ prompt_id: The prompt ID returned from queue_prompt().
+ output_dir: Local directory to save images into.
+ overwrite: If True, overwrite existing files.
+
+ Returns:
+ List of result dicts from download_image(), one per image.
+
+ Raises:
+ RuntimeError: If the prompt is not found or no images are available.
+ """
+ images = list_output_images(base_url, prompt_id)
+
+ if not images:
+ raise RuntimeError(f"No output images found for prompt: {prompt_id}")
+
+ output_d = Path(output_dir)
+ output_d.mkdir(parents=True, exist_ok=True)
+
+ results = []
+ for img in images:
+ filename = img["filename"]
+ dest = str(output_d / filename)
+ result = download_image(
+ base_url=base_url,
+ filename=filename,
+ output_path=dest,
+ subfolder=img.get("subfolder", ""),
+ image_type=img.get("type", "output"),
+ overwrite=overwrite,
+ )
+ results.append(result)
+
+ return results