Issues

• MCP tools with untyped parameter schemas (bare list, Any) are silently dropped from the ToolRegistry

  JetBrains/koog#2326 · Filter by author ponycompute opened on Oct 31, 2026

• Chat Completions ignores the reasoning field in streaming responses

  JetBrains/koog#2310 · Filter by author elonzh opened on Oct 21, 2026

• Chat Completions drops spaces and newlines from streamed reasoning

  JetBrains/koog#2309 · Filter by author elonzh opened on Oct 21, 2026

• GoogleLLMClient serializes empty "toolConfig": {} when no tools are used, causing HTTP 400 (INVALID_ARGUMENT) from Gemini API

  JetBrains/koog#2308 · Filter by author kid1412621 opened on Oct 20, 2026
  2

• Ollama client sends assistant thinking as "\n" instead of the reasoning text

  JetBrains/koog#2306 · Filter by author Sunckist opened on Oct 20, 2026

• OllamaClient.executeStreaming ignores tools, so models cannot call tools while streaming

  JetBrains/koog#2304 · Filter by author Sunckist opened on Oct 19, 2026
  1

• Broader KMP support for Koog’s data models

  JetBrains/koog#2302 · Filter by author mole828 opened on Oct 19, 2026
  1

• singleRunStrategy should process tool results first, then text message.

  JetBrains/koog#2300 · Filter by author akalini opened on Oct 15, 2026
  2

• Streaming LLM calls produce no OpenTelemetry inference spans (tracked in KG-284)

  JetBrains/koog#2299 · Filter by author hanEck opened on Oct 15, 2026

• koog-bedrock-agentcore-runtime: handlerTimeoutMillis and error handling don't cover streaming responses

  JetBrains/koog#2290 · Filter by author ashakirin opened on Oct 2, 2026
  2

• OpenAI Responses streaming drops reasoning items that have encrypted_content but no summary

  JetBrains/koog#2287 · Filter by author FeernandoOFF opened on Sep 30, 2026

• BedrockLLMClient: OpenAI models get no prompt caching, because Bedrock only offers it on the Responses API

  JetBrains/koog#2286 · Filter by author tomasmontielp opened on Sep 30, 2026

• Java streaming flows can stall cancellation and drop chunks

  JetBrains/koog#2280 · Filter by author kyboi opened on Sep 30, 2026

• Java HTTP client decodes individual SSE lines instead of complete data events

  JetBrains/koog#2279 · Filter by author kyboi opened on Sep 30, 2026

• OllamaClient reports a partial token count as the total

  JetBrains/koog#2278 · Filter by author kyboi opened on Sep 30, 2026

• Add typed chat template arguments to OpenAIChatParams

  JetBrains/koog#2277 · Filter by author kyboi opened on Sep 30, 2026

• RetryingLLMClient retries exceptions that wrap cancellation

  JetBrains/koog#2273 · Filter by author kyboi opened on Sep 29, 2026

• Streaming tool calls have inconsistent identity or complete prematurely

  JetBrains/koog#2271 · Filter by author kyboi opened on Sep 29, 2026

• Ktor client ignores request headers instead of overriding defaults

  JetBrains/koog#2269 · Filter by author kyboi opened on Sep 29, 2026

• Skill discovery swallows coroutine cancellation

  JetBrains/koog#2267 · Filter by author kyboi opened on Sep 29, 2026

• Skills YAML prompt changes multiline values and metadata keys

  JetBrains/koog#2265 · Filter by author kyboi opened on Sep 29, 2026

• Skills JSON prompt contains unescaped control characters

  JetBrains/koog#2263 · Filter by author kyboi opened on Sep 29, 2026
  1

• Anthropic client rejects non-null anyOf tool parameters

  JetBrains/koog#2260 · Filter by author kyboi opened on Sep 29, 2026

• Koog memory provider for selected cross-run context

  JetBrains/koog#2259 · Filter by author vivekgupta-memcode opened on Sep 29, 2026

• Deserialization failure when streaming response has no delta.

  JetBrains/koog#2258 · Filter by author jacekgajek opened on Sep 28, 2026