zylon-ai/private-gpt · error · ContentRequestLimitError

Artifact {artifact} exceeds the {self.max_content_depth} dep

Error message

Artifact {artifact} exceeds the {self.max_content_depth} depth limit

What it means

ContentRequestLimitError raised after tree rebuild when max(node.depth) over the artifact's nodes exceeds self.max_content_depth. It guards against pathological nesting (cycles, corrupt parent links, or genuinely deep documents) before yielding the artifact root.

Source

Thrown at private_gpt/server/content/content_service.py:345

                continue
            if len(nodes) > self.max_content_nodes:
                raise ContentRequestLimitError(
                    f"Artifact {artifact} exceeds the {self.max_content_nodes} node limit"
                )

            # Sort nodes by their position in the tree
            nodes = sorted(
                nodes,
                key=lambda n: n.abs_idx,
            )

            # Rebuilt the tree structure
            root_nodes = TreeNode.rebuild_tree(nodes)
            root_node = root_nodes[0] if root_nodes else None
            if not root_node:
                continue
            if max((node.depth for node in nodes), default=0) > self.max_content_depth:
                raise ContentRequestLimitError(
                    f"Artifact {artifact} exceeds the {self.max_content_depth} depth limit"
                )

            yield artifact, root_node

            # Cleanup
            del root_node
            del nodes

    async def retrieve_document_nodes_async(
        self,
        context_filter: ContextFilter,
        include: list[type[NodeType]] | None = None,
        exclude: list[type[NodeType]] | None = None,
        node_ids: list[str] | None = None,
        include_children: bool = True,
        include_ancestors: bool = False,
    ) -> AsyncGenerator[tuple[str, TreeNode]]:

View on GitHub (pinned to 4a030776a3)

Solutions

Inspect the artifact's node depths (node.depth distribution) to confirm whether data is corrupt or legitimately deep.
Re-ingest the source document if depth metadata is corrupt.
Raise max_content_depth for legitimately deep artifacts.
Fix ingestion to cap nesting depth when building nodes.

Example fix

# settings
# before
max_content_depth: int = 20
# after
max_content_depth: int = 64

Defensive patterns

Strategy: validation

Validate before calling

depths = [n.depth for n in nodes]
if max(depths, default=0) > service.max_content_depth:
    flag_artifact_for_repair(artifact)  # corrupt or too deep

Try / catch

try:
    for a, root in service.stream_content(context_filter):
        yield a, root
except ContentRequestLimitError:
    log.error('artifact %s exceeds depth cap — re-ingest source', a)

Prevention

Validate depth metadata at ingestion time
Cap nesting when building the tree from nested sources
Re-ingest sources whose trees exceed the cap instead of raising the limit blindly

When it happens

Trigger: An artifact whose rebuilt tree contains a node deeper than max_content_depth; commonly caused by corrupt abs_idx/parent metadata producing degenerate chains, or deeply nested source structures (nested lists, XML).

Common situations: Ingestion bugs writing wrong depth/parent fields; documents with pathological nesting; max_content_depth default too low for legitimate deep trees; TreeNode.rebuild_tree chaining nodes linearly due to missing parents.

Related errors

AI-assisted analysis of zylon-ai/private-gpt@4a030776a3 (2026-08-15). Data as JSON: /api/errors/fa71d9460e8a7aff. Report an issue: GitHub.