BerriAI/litellm · error · ValueError

Amazon Nova Canvas {task_type} requires a text prompt. Pass

Error message

Amazon Nova Canvas {task_type} requires a text prompt. Pass a non-empty `prompt` in your request.

What it means

Bedrock's Nova Canvas INPAINTING and OUTPAINTING tasks require a non-empty text prompt (it drives the mask semantics / area description). After popping taskType, the transformer checks prompt and raises ValueError when it is None or '' for those two task types; other tasks fall back to a single-space text.

Source

Thrown at litellm/llms/bedrock/image_edit/amazon_nova_canvas_image_edit_transformation.py:352

        if width is not None:
            image_generation_config["width"] = width
        if height is not None:
            image_generation_config["height"] = height
        if number_of_images is not None:
            image_generation_config["numberOfImages"] = number_of_images
        if quality is not None:
            image_generation_config["quality"] = quality
        if cfg_scale is not None:
            image_generation_config["cfgScale"] = cfg_scale
        if seed is not None:
            image_generation_config["seed"] = seed

        task_type: Final = op.pop("taskType", None)
        if (prompt is None or prompt == "") and task_type in (
            "INPAINTING",
            "OUTPAINTING",
        ):
            raise ValueError(
                f"Amazon Nova Canvas {task_type} requires a text prompt. Pass a non-empty `prompt` in your request."
            )
        text: Final = prompt if prompt is not None and prompt != "" else " "
        negative_text: Final = op.pop("negativeText", None)
        similarity_strength: Final = op.pop("similarityStrength", None)
        mask_prompt: Final = op.pop("maskPrompt", None)
        out_painting_mode: Final = op.pop("outPaintingMode", None)

        body: Final = _nova_canvas_task_body(
            image_b64=image_b64,
            mask_b64=mask_b64,
            text=text,
            negative_text=negative_text,
            similarity_strength=similarity_strength,
            task_type=task_type,
            mask_prompt=mask_prompt,
            out_painting_mode=out_painting_mode,
        )

View on GitHub (pinned to 6c2dcb801b)

Solutions

Supply a meaningful non-empty prompt describing what to paint into the masked area (or, for outpainting, how to extend the scene).
Double-check the field name is 'prompt' in the multipart request.
If you genuinely want no text, use IMAGE_VARIATION or BACKGROUND_REMOVAL, which tolerate a default text.

Example fix

# before
litellm.image_edit(model="bedrock/nova-canvas", image=img,
                  taskType="INPAINTING", maskPrompt="sky", prompt="")  # ValueError
# after
litellm.image_edit(model="bedrock/nova-canvas", image=img,
                  taskType="INPAINTING", maskPrompt="sky",
                  prompt="fill with blue sky")

Defensive patterns

Strategy: validation

Validate before calling

def validate_prompt_for_task(prompt: str | None, task_type: str | None) -> None:
    if task_type in ("INPAINTING", "OUTPAINTING") and (prompt is None or prompt == ""):
        raise ValueError(f"{task_type} requires a non-empty prompt")

Prevention

Require non-empty prompt strings in your form/API layer for edit endpoints.
Treat empty prompt as a client bug, not a default.
Document per-task prompt requirements next to taskType constants.

When it happens

Trigger: images/edits with taskType='INPAINTING' or 'OUTPAINTING' and prompt omitted, None, or empty string; prompt stripped by upstream validation before reaching litellm.

Common situations: Mask-only inpainting requests carried over from OpenAI (where prompt is also required but sometimes faked as ''); building edit requests programmatically where prompt defaults to None; prompt passed under a different key name.

Related errors

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/fc73a68afe95b709. Report an issue: GitHub.