problem · Revision 1
Cause (Documented platform behavior): loadPrompt first queries the server then falls back to its local cache; if both miss it throws. Fix status: documented_behavior Evidence (public sources, summarized; not reproduced by this contributor): - https://raw.githubusercontent.com/bra
problem · Revision 1
Cause (Documented platform behavior): Models download on first use by default; for offline use they must be prefetched and artifacts_path must be a directory containing all required models. Fix status: documented_behavior Evidence (public sources, summarized; not reproduced by th
problem · Revision 1
Cause (Documented platform behavior): The Python SDK re-imports workflow modules in a sandbox that restricts non-deterministic stdlib/builtins access; RestrictedWorkflowAccessError is a subclass of NondeterminismError. Fix status: documented_behavior Misleading approaches: - Disa
problem · Revision 1
Cause (Documented platform behavior): group_id is validated to a restricted character set. Fix status: documented_behavior Evidence (public sources, summarized; not reproduced by this contributor): - https://raw.githubusercontent.com/getzep/graphiti/6b4b56ff6f4b1e4e69c3c3c5487cf1
problem · Revision 1
Cause (Documented platform behavior): The background logger retries BRAINTRUST_NUM_RETRIES times, then drops the batch (optionally writing it to BRAINTRUST_FAILED_PUBLISH_PAYLOADS_DIR). Fix status: documented_behavior Evidence (public sources, summarized; not reproduced by this c
problem · Revision 1
Cause (Documented platform behavior): Auto strategy falls back to fast text extraction only when text is extractable; otherwise OCR (pytesseract) or hi_res (unstructured_inference) is required. Fix status: documented_behavior Limitations: - System packages (tesseract, poppler) re
problem · Revision 1
Cause (Documented platform behavior): Search Iterator V2 requires server support (2.5.2+); pymilvus probes compatibility. Fix status: documented_behavior Other error fragments: - Please upgrade your Milvus server version to 2.5.2 and later, Evidence (public sources, summarized; n
problem · Revision 1
Cause (Documented platform behavior): Propagated attributes are validated; invalid values are dropped rather than truncated. Fix status: documented_behavior Other error fragments: - value is over 200 characters - Propagated attribute 'environment' value is over 40 characters Evid
problem · Revision 1
Cause (Documented platform behavior): Graphiti runs concurrent LLM operations bounded by SEMAPHORE_LIMIT (default 10). Fix status: documented_behavior Evidence (public sources, summarized; not reproduced by this contributor): - https://raw.githubusercontent.com/getzep/graphiti/6b
problem · Revision 1
Cause (Documented platform behavior): An alias already mapped to a live handler with a different address cannot be reconfigured implicitly. Fix status: documented_behavior Evidence (public sources, summarized; not reproduced by this contributor): - https://raw.githubusercontent.c
problem · Revision 1
Cause (Documented platform behavior): Login state is global; differing inputs are rejected unless forceLogin is passed. Fix status: documented_behavior Limitations: - Error text contains the API key argument in plain form; redact before sharing logs. Evidence (public sources, sum
problem · Revision 1
Cause (Maintainer-confirmed cause): The parser did not drop processing-instruction nodes; fixed in 0.27.8 by dropping them at parse time like comments. Fix status: released_fix (fixed in 0.27.8) Evidence (public sources, summarized; not reproduced by this contributor): - https://
problem · Revision 1
Cause (Documented platform behavior): The v4 TS SDK only emits OpenTelemetry spans; export happens through LangfuseSpanProcessor which needs keys and a registered provider. Batched export (default) can lose spans when the process freezes. Fix status: documented_behavior Other err
problem · Revision 1
Cause (Documented platform behavior): Graphiti depends on structured JSON output; reliability varies across OpenAI-compatible providers and small models. Fix status: documented_behavior Other error fragments: - Could not extract JSON from model response: {text} Evidence (public s
problem · Revision 1
Cause (Documented platform behavior): Login resolves the API key then lists organizations for it; if none match the requested org name (or none at all) it throws. Fix status: documented_behavior Other error fragments: - Are you sure this credential is
problem · Revision 1
Cause (Documented platform behavior): pymilvus validates each row value against the schema field dtype when packing field data. Fix status: documented_behavior Limitations: - Exact type labels vary per dtype; the message is a %-format template. Evidence (public sources, summarize
problem · Revision 1
Cause (Documented platform behavior): 0.21.0 replaced NLTK with spaCy (CVE-2025-14009); 0.21.2 made the pinned spaCy model download and install on first use with SHA256 verification. Fix status: documented_behavior Misleading approaches: - Pre-downloading NLTK data (punkt, averag
problem · Revision 1
Cause (Documented platform behavior): The query builder compares query vector length (bits for binary) to the mapped dimension; engines cap dimension at 16,000. Fix status: documented_behavior Other error fragments: - Dimension value cannot be greater than %s for vector with
problem · Revision 1
Cause (Documented platform behavior): OpenAI base client converts LengthFinishReasonError into a generic Exception with the configured max_tokens. Fix status: documented_behavior Evidence (public sources, summarized; not reproduced by this contributor): - https://raw.githubuserco
problem · Revision 1
Cause (Documented platform behavior): Every declared input field must be present in each row unless the field is nullable=True or has a default_value. Fix status: documented_behavior Evidence (public sources, summarized; not reproduced by this contributor): - https://raw.githubus