problem · Revision 1
Finding: In an external report against LiteLLM v1.99.1 (ghcr.io/berriai/litellm:main-stable, image revision 10f4033437df30b91b5dbf2b64711d0a8683fc52, Postgres 16, Terraform v1.10.5), updating a virtual key that has no team through the Terraform provider fails with HTTP 500 and th
problem · Revision 1
Finding: In vLLM RFC #59502 (modulewise weight reload), an external reviewer reports three possible correctness gaps in the author's prototype branch. They concern RL weight updates through start_weight_update, receive_weights and finish_weight_update, and the version is UNKNOWN
problem · Revision 1
Finding: In an external report, a crewAI conversational Flow served over AG-UI with ag-ui-crewai (conversational=True) reportedly shows duplicated or out-of-order chat messages in CopilotKit/AG-UI frontends, because historical messages get a new random id on every MESSAGES_SNAPSH
problem · Revision 1
Finding: In an external report against LiteLLM v1.106.0 (https://github.com/BerriAI/litellm/issues/45558), the Responses API bridge (litellm.aresponses with reasoning effort set) is said to place provider reasoning text in a different reasoning-item field depending on streaming m
problem · Revision 1
Finding: In an external report against pnpm 12.8.0, repeated `pnpm dedupe --lockfile-only` runs on a two-project workspace reportedly flip one peer-variant key of `@react-native-firebase/app@23.8.2` between a hashed suffix and an expanded suffix. As a result `pnpm dedupe --check`
problem · Revision 1
Finding: In an externally reported case, streamText with ai 7.0.128, @ai-sdk/openai-compatible 3.0.63 and @ai-sdk/provider-utils 5.0.54 (Node 22+) mishandles parallel calls of the same tool when an OpenAI-compatible stream reuses or omits tool_calls[].index. Source: https://githu
problem · Revision 1
Finding: In vercel/ai issue #22083, the reporter says that with ai 7.0.128 (also seen on 7.0.127 in production), a Safari/WebKit network failure during a streaming useChat/AbstractChat request surfaces as TypeError 'Load failed'. It is not classified as a disconnect: onFinish
problem · Revision 1
Finding: In an external report on pnpm issue 16354, a fresh pnpm install --lockfile-only with injectWorkspacePackages: true records a workspace peer dependency as a peer-suffixed file: entry under importers in pnpm 12.0.0 through 12.8.1, where pnpm 11.28.2 records link: (https://
problem · Revision 1
Finding: In an external report against crewAI main (HEAD 42ae4bf, Python 3.12, macOS), create_model_from_schema reportedly drops field constraints such as pattern and minimum when a JSON Schema property uses the list-form nullable type, e.g. ["string","null"] or ["integer","null"
problem · Revision 1
Finding: In one external report (cloudflare/workers-sdk#16052, https://github.com/cloudflare/workers-sdk/issues/16052), with wrangler 4.135.0 and a Worker Previews open-beta setup (compatibility date 2026-09-20), a fetch() from a Worker to a Preview served on a custom domain retu
problem · Revision 1
Finding: In langgraph issue #2555, a graph whose state is a Pydantic model with Field(alias='bar') reportedly fails with a validation error when invoked with the alias key {'bar': 'hello'}; the langgraph and pydantic versions are not stated (UNKNOWN). Source: https://github.com/l
problem · Revision 1
Finding: In the MCP filesystem server's edit_file tool (src/filesystem/lib.ts), the issue reports that when exact matching fails and the whitespace-tolerant line matcher is used, replacement text does not keep the file's indentation. This is reported for the legacy-era (2025-11-2
problem · Revision 2
Finding: In Gemini CLI built from commit fb972b2f87 (0.65.0-nightly.20261006), an external report says write_file and replace can fail with ENAMETOOLONG when the destination file name is about 215–255 bytes, even though the name itself is legal. This is an external
problem · Revision 2
Finding: In an external report, running `vitest run` (vitest 5.0.2, vite 8.3.1, wrangler 4.144.0, Node 24.18.0, pnpm 12.6.0, macOS arm64) in a project whose Vite config includes `cloudflare()` from @cloudflare/vite-plugin fails at startup with "There is already a server associate
problem · Revision 2
Finding: In a pnpm workspace on pnpm 12.0.0 through 12.8.2 (reported on macOS, Node v24.14.1), an unrelated manifest change in one project reportedly moved another project's direct dependency to a lower version, even though its specifier was unchanged and its
problem · Revision 2
Finding: In a pnpm workspace, an external report says `pnpm install --frozen-lockfile` fails with ERR_PNPM_OUTDATED_LOCKFILE on pnpm 12.7.0, 12.8.0 and 12.8.1 when an injected workspace package (`lib`) declares an optional peer via `peerDependenciesMeta` and a sibling importer (`
problem · Revision 2
Finding: In an external report against langgraph 1.2.10 (Windows, Python 3.14.7), constructing a StateGraph with an Annotated reducer such as set.union is reported to raise a ValueError. The reporter's traceback ends with: ValueError: Invalid reducer signature. Expected (a, b) ->
problem · Revision 2
Finding: In an open workers-sdk report, a cloudflare.config.ts Worker that binds its own Durable Object class (binding worker field set to the Worker's own name) fails under @cloudflare/vitest-plugin with experimental newConfig enabled, and workerd reportedly refuses to start bef
problem · Revision 2
Finding: In @modelcontextprotocol/sdk 1.29.0 (and the v1.x and main code at the commits cited in the issue), ReadBuffer.readMessage() in shared/stdio.ts advances with subarray(index + 1). When the newline is the final byte, this leaves an empty Buffer view that still
problem · Revision 2
Finding: In Wrangler (reported as wrangler@latest and reproduced on main, exact released version unknown), running `wrangler dev` with the CLOUDFLARE_ENV variable set applies the matching config environment but is reported not to load `.dev.vars.<env>` or `.env.<env>`. Passing `-