Severity
Reports at this severity, newest stored first.
When a model returns invalid structured output, ToolStrategy's default handle_errors=True causes infinite retries without any attempt limit, leading to context window exhaustion or recursion limit errors. A retry budget should be introduced to fail gracefully after a maximum number of attempts.
Turbopack rejects SVG images that have a width attribute and a viewBox but no height, causing dev and build failures, while webpack handles the same SVG correctly by deriving height from the viewBox aspect ratio.
The `TypeSafeClassifier` input schema generation fails with `PydanticUserError` because the private recursive `_StateValue` type alias is not resolvable in the dynamically created `TypeSafeClassifierInput` model namespace. This breaks `get_input_jsonschema()`, `as_tool()`, schemaful graph serialization, and Runnable composition schema introspection.
ChatAnthropic in langchain-anthropic 1.6.1 sends top_k=10 to claude-opus-5, which Anthropic has deprecated for Opus 4.7 and later, resulting in a 400 BadRequestError. The client should validate locally and reject unsupported parameters before making a network request.
Next.js build-time Instant Validation for Cache Components does not populate staged fallback params in the synthetic RequestStore, causing false negatives for invalid param access outside Suspense boundaries in dynamic routes with PPR.
The LLMToolSelectorMiddleware only passes the last user message to the selector model, discarding prior conversation context. This causes vague follow-up queries (e.g., 'do it') to be misinterpreted, selecting incorrect tools.
When using Router.replace or Router.push with a href that includes query parameters followed by a hash (e.g. /page?test=1#hash) and an as path that also includes the same hash, the getInitialProps method is not re-invoked on subsequent query changes. This breaks data fetching for masked-query patterns where query is hidden from the browser URL but hash must remain visible.
When a request receives a stale cached entry from a "use cache" function, the background revalidation incorrectly replaces the saved entry in the Resume Data Cache with the pending result of the regeneration. This leads to dev-time "uncached data" errors on validation renders and in production to runtime prefetch/RSC responses missing or mixing cache versions when using cacheComponents and partialPrefetching.
With App Router, dynamic pages using revalidate=0 or no-store fetch and a global loading.tsx get stuck on the loading fallback when JavaScript is disabled, making content invisible to crawlers and harming SEO.
Clearing a Storage policy's USING or WITH CHECK expression in the Studio dashboard shows a success toast but does not persist the change because the payload omits undefined fields, leaving the old expression active and potentially causing security misconfigurations.
ChatAnthropic allows forced tool_choice and thinking={"type":"disabled"} to reach the Anthropic API for claude-opus-5-5, causing 400 invalid-request errors. bind_tools(), call-time kwargs, model_kwargs, and create_agent() with structured-output tools are affected.
TypeSafeClassifier from langchain-typesafe declares is_lc_serializable() True and dumpd() succeeds, but load() fails with both allowed_objects='all' and allowed_objects=[TypeSafeClassifier]. The root cause is that the serialized ID ('langchain', 'classifiers', 'typesafe', 'TypeSafeClassifier') is not present in langchain-core's deserialization mapping and the 'langchain_typesafe' namespace is not in the default valid namespaces.
ChatAnthropic incorrectly routes the documented browser_toolset_20260801 client toolset through JSON-schema conversion, raising ValueError. Non-null toolset_name on tool_use blocks is lost in multi-turn requests, and ordinary tool calls expose toolset_name=None.
Importing a CSV file with a geometry column where individual rows have a large number of points (over approximately 1000) causes the Supabase dashboard import process to hang. This prevents data import for such rows and requires manual splitting or exclusion.
langchain-typesafe's TypeSafeClassifier does not validate that the 'questions' mapping is non-empty before making an HTTP request, causing invalid API calls and a regression from previous Pydantic-based validation.
When ContextEditingMiddleware uses token_count_method="model" through the async method awrap_model_call, it directly invokes the synchronous model.get_num_tokens_from_messages on the event-loop thread. For models like ChatAnthropic that perform a synchronous HTTP request for token counting, this blocks the entire event loop for the duration of each count, causing severe latency spikes and starving other async tasks.
Running `npm audit` on a project created with Next.js 15.5.23 reports high severity vulnerabilities from transitive dependencies `postcss` and `sharp`. The Next.js package pins outdated versions of these packages, requiring a dependency bump to secure patch releases.
Next.js 15 preloads all dynamic imports inside client components, causing both conditional chunks to be sent to the client, increasing initial JavaScript size and TBT compared to Next.js 14. The issue requires a user-land workaround until an upstream fix is released.
On supabase.com, opening and closing a navbar mega-menu leaves a permanently mounted container (forceMount) that overlays hero content and intercepts pointer events, preventing text selection and CTA clicks.
Setting `"type": "commonjs"` in a Next.js project using Turbopack causes build and dev failures because Turbopack enforces the declared module format on App Router source files (which are written as ESM) and on generated `.next-internal` server-action entries, even though Webpack ignores this field. The issue affects all App Router files and internal virtual modules, not just the `'use client'` proxy previously fixed.
langchain-openai fails to classify errors embedded in HTTP 200 responses from OpenAI-compatible gateways. Non-streaming invoke raises a bare ValueError, streaming raises an unclassified openai.APIError, and Responses API raises ValueError. This prevents callers from accessing is_retryable and other ModelError attributes, breaking retry logic and potentially causing silent degradation.
The `_convert_to_v1_from_converse` translator in langchain-core drops redacted reasoning content blocks (e.g., from GPT-6 Sol/Luna/Astra via Bedrock), causing loss of reasoning context in multi-turn conversations when using `output_version="v1"`. This leads to the next turn not sending back the reasoning, potentially degrading model performance.
The langchain-mcp-adapters library discards structuredContent when an MCP tool returns isError=True. It raises _MCPToolExecutionError before reading the field and never attaches it to the resulting error ToolMessage, unlike the success path. This causes loss of machine-readable error payloads and leaves only a placeholder text when content is empty.
In langchain-core's _compat_bridge._iter_protocol_blocks, tool_call content blocks lacking an explicit index are keyed by ("__lc_no_index__", block_type, position_in_chunk). When Gemini streams parallel function calls as separate chunks each containing one complete tool_call block at position 0, the keys collide. _accumulate then overwrites the earlier block with the later one, silently dropping the first tool call. This leads to incorrect model behavior and, with Gemini 3, the lost call may carry the thought_signature, causing upstream validation issues or silent execution of only one of the requested tools.
When a dict is passed to EvaluationResult.feedback_config, the dataclass __post_init__ coerces it to FeedbackConfig, which ignores unknown fields. This causes silent data loss and contradicts the Union[FeedbackConfig, dict] type hint.
`InMemoryVectorStore.add_documents` and `aadd_documents` silently drop documents when the embedding model returns fewer vectors than the number of input documents. The root cause is the use of `zip(..., strict=False)` in the underlying `add_texts`/`aadd_texts` methods. A `ValueError` should be raised immediately to prevent silent data loss.
The `add_texts` method in `langchain-core` does not raise a `ValueError` when the number of provided IDs does not match the number of texts, despite the docstring stating otherwise. This leads to silent data loss as extra texts are discarded via `zip(strict=False)`. A regression from commit `6dd9f053e3` removed explicit validation.
Merging two AddableDict instances with different value types for the same key silently discards the left-hand value and replaces it with the right-hand value, potentially causing hidden data loss in LCEL stream aggregation.
BasePromptTemplate._validate_input uses isinstance(inner_input, dict) instead of collections.abc.Mapping, causing standard library mappings that are not dict subclasses to be mishandled. Single-variable templates silently render the mapping's repr into the prompt, while multi-variable templates raise a misleading TypeError. All template formats and both sync/async paths are affected. Fix is to check Mapping and convert to dict.
Using create_agent with response_format=ToolStrategy and a model returning an AIMessage without tool_calls results in a response missing the structured_response key instead of raising an error (handle_errors=False) or retrying (handle_errors=True), causing silent downstream failures.