langchain · Issues· 536 open

core: `InMemoryVectorStore` raises `NotImplementedError` for relevance-score search and the `similarity_score_threshold` retriever

InMemoryVectorStore.delete(None) does not delete all documents

`HTMLHeaderTextSplitter` raises `ValueError` on init for the non-heading tags its splitter already supports

core: content_blocks raises TypeError on a v0 multimodal block that carries an id

TimeWeightedVectorStoreRetriever keeps documents in memory after vector store insertion fails

RunnableAssign leaves mapper work running after passthrough input validation fails

core: mustache inverted section `{{^key}}` nested inside a `{{#key}}` list section misrenders or raises `IndexError`

core: `AsyncBaseTracer` drops `name` in `on_tool_start` and `response` in `on_llm_error`, diverging from `BaseTracer`

Blob leaks a UTF-8 BOM into as_string()

ExperimentalMarkdownSyntaxTextSplitter mutates results returned by previous split_text calls

QAWithSourcesChain removes docs from caller-provided inputs

UsageMetadataCallbackHandler silently drops usage_metadata when model_name is missing

core: `finalize_tool_call_chunk` coerces `id=None` to `""`, collapsing parallel id-less tool calls

langchain 1.4.0: any after_model middleware using documented `jump_to="tools"` bypasses the HITL pending_tool_calls gate and re-executes rejected tool calls

core: docstrings shouldn't inherit from parents