메뉴
Use when local Copilot/Qwen tool calls fail, malformed tool JSON appears, tools are skipped, or LLM Gateway agent mode is flaky.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when local Copilot/Qwen/LLM Gateway hits prompt bloat, context limits, slow prefill, timeouts, or 24GB GPU token-budget issues.
Use when local Qwen, llama-server, or ik_llama.cpp dies, stalls, times out, OOMs, cancels, or needs log-based failure diagnosis.
Use when a local-agent coding task needs minimal repo context, targeted file discovery, compact search, or a small implementation brief.
Use when a local Copilot/Qwen chat is too long and needs compaction, checkpointing, restart, or a handoff summary.
Use when local Qwen/Copilot should delegate isolated research, review, or summarization to subagents without bloating parent context.
| name | local-tool-reliability |
| description | Use when local Copilot/Qwen tool calls fail, malformed tool JSON appears, tools are skipped, or LLM Gateway agent mode is flaky. |
| argument-hint | [tool-call symptom or error message] |
Debug and stabilize local-model tool calling before changing application code.
llama-server is running and /v1/models responds.github.copilot.llm-gateway.agentTemperature: 0.0 for tool mode.parallelToolCalling: false for Qwen.enableToolCalling: true unless isolating a basic chat failure.requestTimeout for large local prompts.--jinja).preserve_thinking: true).qwen3_coder tool parser where available.q8_0 when reliability matters.Return: