problem · Revision 3
## Observed symptom [BUG] MCP client tears down the built-in IDE MCP Server session: No ContextAccessor for contextType reactor.util.context.Context1 (regression from #86) The built-in MCP client cannot list tools for the IDE's own built-in MCP server (`IDE MCP Server`): the
problem · Revision 2
## Observed symptom [BUG] Structured output: with the Gemini provider, the answer is published as a set_model_response call before validation - [X] I have searched the [existing issues](https://github.com/kagent-dev/kagent/issues) to avoid creating a duplicate - [X] By submitting
problem · Revision 2
## Observed symptom bug(core): 400 Bad Request "Requests ending with a model turn are not supported" when reading image files via ReadFile tool When Gemini CLI attempts to inspect image files (e.g. `.png`) using the built-in `ReadFile` tool during an
problem · Revision 3
## Observed symptom 1.x: genai-prices 0.1.x silently zeroes all token usage — cap it below 0.1 in 1.107.x - [x] I'm using the [latest version](https://github.com/pydantic/pydantic-ai/releases/latest) of Pydantic AI - [x] I've searched for my issue in [the issue tracker](https://g
problem · Revision 2
## Observed symptom [bug] anthropic: early exit from a stream records the SDK's placeholder tool arguments ({}) as the final tool call When the application stops a streamed Anthropic response partway (here: breaking out of `client.messages.stream(...)` in the middle of
problem · Revision 2
## Observed symptom [BUG] Anthropic provider fails on every call with anthropic>=1.0: `Messages.create() got an unexpected keyword argument 'temperature'` **Bug Description** anthropic 1.0.0 removed `temperature`, `top_p` and `top_k` from the signature of `Messages.create`. The S
problem · Revision 2
## Observed symptom [Bug]: claude-opus-5-5 requests ending in an assistant (prefill) message fail with Anthropic 400 "does not support assistant message prefill" Is there an existing issue for the same bug? - [x] I have searched existing issues and this
problem · Revision 2
## Observed symptom AI Gateway: Anthropic system-message placement breaks Codex continuation and compaction Codex sessions using Claude Opus through AI Gateway fail when a refreshed developer instruction follows an assistant's final answer. The gateway converts the update to an i
problem · Revision 3
## Observed symptom [Bug]: Validation errors split a single tool-call response in SDK conversation history Is there an existing issue for the same bug? - [x] I have searched existing issues and this is not a duplicate. ## Environment GitHub
problem · Revision 2
## Observed symptom [Bug] gpt-oss: `tool_choice: "required"` returns raw harmony control tokens as assistant content - [x] I searched related issues but found no solution. - [x] The bug persists in the latest version. - [x] Issues without environment info
problem · Revision 3
## Observed symptom [BUG] v4.7.10: video_generate() default runway/gen3a_turbo invalid — fails before auth [BUG] v4.7.10: `praisonai.capabilities.video_generate()` default `runway/gen3a_turbo` still invalid — fails before provider authentication Metadata --- Executive summary The
problem · Revision 3
## Observed symptom [Bug] Design: LLM stream ReadError is masked as "node agent returned empty output" Paired: [GitHub #7657](https://github.com/openJiuwen-ai/jiuwenswarm/issues/7657) ↔ [GitCode #5001](https://gitcode.com/openJiuwen/jiuwenswarm/issues/5001) Summary In Designer mo
problem · Revision 2
## Observed symptom bug(llms/openai): the OpenRouter models fallback option makes every LLM call raise TypeError With `OPENROUTER_API_KEY` set and the documented `models` fallback option configured (`docs/components/llms/config.mdx`), every OpenAI LLM call raises `TypeError` befo
problem · Revision 2
## Observed symptom wait 60000 fails with MCP error -32001: Request timed out `wait 60000` fails with `MCP error -32001: Request timed out` after about a minute, even though the page is fine. Observed on chrome-devtools-axi 0.1.35, macOS 15.6.1, isolated
problem · Revision 2
## Observed symptom MCP OAuth: Figma token exchange fails with "Failed to parse server response" (distinct from #4870 / #4906) After the `-32601`/`server/discover` fatal-failure bug (#4870) was fixed in 1.0.87-0, the Figma remote MCP server (`https://mcp.figma.com/mcp`) now progr
problem · Revision 2
## Observed symptom Remote MCP: OAuth fails with Perplexity (trailing slash in resource parameter) Remote MCP can't be connected to Perplexity as a custom connector (`https://mcp.desktopcommander.app/mcp`, OAuth, Streamable HTTP). Perplexity sends `resource=https://mcp.desktopcom
problem · Revision 2
## Observed symptom ChatGPT Desktop: MCP OAuth tool call hangs after refresh token returns invalid_grant, while ChatGPT Web reauthenticates successfully What version of the Codex App are you using (From “About Codex” dialog)? N/A — reproduced in ChatGPT Desktop (regular
problem · Revision 3
## Observed symptom [BUG]: Gateway OAuth callback rejects authorization codes longer than 2048 chars (HTTP 422) `GET /oauth/callback` returns HTTP 422 when `code` is longer than 2048 characters. The request fails in query validation, before the token exchange. ContextForge then
problem · Revision 2
## Observed symptom Windows: two Python entrypoints compute conflicting dependency stamps -> endless lazy-install ping-pong (~11 min rebuild each time) ## Environment GitHub issue NousResearch/hermes-agent #125066; state=open ## Trigger / context Windows: two Python entrypoints c
problem · Revision 3
## Observed symptom A2A: outbound text parts advertise mediaType, which strict A2A 1.0 peers reject A2A: outbound text parts advertise `mediaType`, which strict A2A 1.0 peers reject `a2a_call` fails against an A2A 1.0 server that validates parts on member presence.