langchain-ai/langchain · error · ValueError

Unrecognized content block at messages[{i}].content[{j}] has

Error message

Unrecognized content block at messages[{i}].content[{j}] has 'type': 'guard_content' but does not have a messages[{i}].content[{j}]['guard_content']['text'] key. Full content block:

{block}

What it means

The converter recognizes a `"guard_content"` dialect (used for LanGuardian-style guarded text): the block must contain `guard_content` and that nested dict must contain `text` (either a string or `{"text": ...}`). If either is missing, the guard text cannot be extracted and the block is rejected with this error. The check fires both when `type == "guard_content"` lacks the key entirely and when the nested `text` is absent.

Source

Thrown at libs/core/langchain_core/messages/utils.py:1911

                    content.append(
                        {
                            "type": "text",
                            "text": json.dumps(block["json"]),
                        }
                    )
                elif (block.get("type") == "guard_content") or "guard_content" in block:
                    if (
                        "guard_content" not in block
                        or "text" not in block["guard_content"]
                    ):
                        msg = (
                            f"Unrecognized content block at "
                            f"messages[{i}].content[{j}] has 'type': "
                            f"'guard_content' but does not have a "
                            f"messages[{i}].content[{j}]['guard_content']['text'] "
                            f"key. Full content block:\n\n{block}"
                        )
                        raise ValueError(msg)
                    text = block["guard_content"]["text"]
                    if isinstance(text, dict):
                        text = text["text"]
                    content.append({"type": "text", "text": text})
                # VertexAI format
                elif block.get("type") == "media":
                    if missing := [k for k in ("mime_type", "data") if k not in block]:
                        err = (
                            f"Unrecognized content block at "
                            f"messages[{i}].content[{j}] has 'type': "
                            f"'media' but does not have key(s) {missing}. Full "
                            f"content block:\n\n{block}"
                        )
                        raise ValueError(err)
                    if "image" not in block["mime_type"]:
                        err = (
                            f"OpenAI messages can only support text and image data."
                            f" Received content block with media of type:"

View on GitHub (pinned to e32fa9a52e)

Solutions

1. Emit the full shape: `{"type": "guard_content", "guard_content": {"type": "text", "text": "..."}}` (nested `text` as a plain string also works).
2. If moderation is not needed, remove the block rather than leaving a stub.
3. Pin/verify the middleware version whose guard-block schema matches what `langchain-core` expects.

Example fix

// before
{"type": "guard_content", "guard_content": {"type": "text"}}

// after
{"type": "guard_content", "guard_content": {"type": "text", "text": "guarded payload"}}

def valid_guard_block(b: dict) -> bool:
    gc = b.get("guard_content")
    return isinstance(gc, dict) and (
        isinstance(gc.get("text"), str)
        or isinstance(gc.get("text"), dict) and "text" in gc["text"]
    )

blocks = [b for b in blocks if b.get("type") != "guard_content" or valid_guard_block(b)]

def is_guard_content_block(b: dict) -> bool:
    return (
        (b.get("type") == "guard_content" or "guard_content" in b)
        and isinstance(b.get("guard_content"), dict)
        and "text" in b["guard_content"]
    )

When it happens

Trigger: `{"type": "guard_content"}` with no `guard_content` key; `{"guard_content": {"type": "text"}}` where the nested payload omits `text`; guard blocks produced by a middleware version with a different schema.

Common situations: Integrating content-moderation middleware that emits `guard_content` blocks; upgrading the middleware after its schema changed; hand-crafting guard blocks in tests.

Related errors

• Unrecognized content block at messages[{i}].content[{j}] has
• Unrecognized content block at messages[{i}].content[{j}] has
• Unrecognized content block at messages[{i}].content[{j}] has
• Unrecognized content block at messages[{i}].content[{j}] has
• Unrecognized content block at messages[{i}].content[{j}] doe

AI-assisted analysis of langchain-ai/langchain@e32fa9a52e (2026-08-14). Data as JSON: /api/errors/931468b4b538a282. Report an issue: GitHub.