# つれづれなる Agent OPS https://llm-lab.dev ## Overview Site Name: つれづれなる Agent OPS (Tsurezure Agent OPS) Language: Japanese (primary) / English (translated) Format: Static blog built with Astro + MDX Hosting: Cloudflare Pages ### Description (日本語) LLMOps、AI、MCP、AG-UI、Langfuse などの実験ログ。 「実際に動かして、壊れ方まで記録する」を方針とした実装ノートです。 AgentOps/LLMOps のニッチ領域で網羅的な実装ログを目指しています。 ### Description (English) Practical notes on LLMOps, AI agents, MCP, Langfuse, Cloudflare, and related experiments. A blog that records not just successes, but how things break. Aims to be a comprehensive implementation log in the AgentOps/LLMOps niche. ## Main Sections - 技術メモ (Tech Notes): 18 articles - AgentOps: 7 articles - Flue: 7 articles - AIDD: 4 articles - eve: 4 articles - 設計 (Design): 3 articles - Clickhouse: 2 articles - 実装検証 (Implementation): 2 articles - 運用観測: 2 articles - Hermes Agent: 2 articles - Sakana Fugu: 2 articles - Cloudflare: 2 articles - Generative UI: 1 article - 技術検証 (Tech Validation): 1 article ## Popular Articles Top 15 recent articles: - SWE-1.7の自己圧縮から考えるAIDDの再開メモ設計 https://llm-lab.dev/posts/aidd-self-compaction-resume-check/ https://llm-lab.dev/en/posts/aidd-self-compaction-resume-check/ - LangfuseのLLMトレースをClickHouseで長期分析する https://llm-lab.dev/posts/langfuse-clickhouse-trace-analytics/ https://llm-lab.dev/en/posts/langfuse-clickhouse-trace-analytics/ - Cloudflare PagesとWorkersとAI GatewayでAIチャットBotの土台を作る https://llm-lab.dev/posts/cloudflare-pages-workers-ai-gateway-chatbot/ https://llm-lab.dev/en/posts/cloudflare-pages-workers-ai-gateway-chatbot/ - ループは回った。でも何が効いたのか分からない https://llm-lab.dev/posts/llm-loop-engineering-langfuse-observability/ https://llm-lab.dev/en/posts/llm-loop-engineering-langfuse-observability/ - Cloudflare AI GatewayのpatchLogでLLMの評価ログを育てる https://llm-lab.dev/posts/cloudflare-ai-gateway-patch-log-feedback-loop/ https://llm-lab.dev/en/posts/cloudflare-ai-gateway-patch-log-feedback-loop/ - FlueのAgentはどこで失敗するのか Skill実行と構造化出力を切り分ける https://llm-lab.dev/posts/flue-skill-structured-output-comparison/ https://llm-lab.dev/en/posts/flue-skill-structured-output-comparison/ - Eveのeval失敗ログをLangfuse Datasetに残して評価条件を改善する https://llm-lab.dev/posts/vercel-eve-eval-langfuse-dataset/ https://llm-lab.dev/en/posts/vercel-eve-eval-langfuse-dataset/ - ルーブリックとEvalは最初から正しく作れないので、失敗ログで育てる https://llm-lab.dev/posts/llm-loop-engineering-rubric-eval/ https://llm-lab.dev/en/posts/llm-loop-engineering-rubric-eval/ - Langfuse Assistantは運用調査の入口になるかをAPIの正解データで検証する https://llm-lab.dev/posts/langfuse-assistant-public-beta-verification/ https://llm-lab.dev/en/posts/langfuse-assistant-public-beta-verification/ - 個人ブログをAI Agentに読ませる llms.txtとMCP ServerでAIOを整える https://llm-lab.dev/posts/blog-aio-llms-mcp/ https://llm-lab.dev/en/posts/blog-aio-llms-mcp/ - APIで最小ループを組むと、プロンプト改善ではなく停止条件の設計になる https://llm-lab.dev/posts/llm-loop-engineering-minimal-api/ https://llm-lab.dev/en/posts/llm-loop-engineering-minimal-api/ - FlueでOpenAI互換APIを使うときはmodel specifierを先に疑う https://llm-lab.dev/posts/flue-openai-compatible-provider-note/ https://llm-lab.dev/en/posts/flue-openai-compatible-provider-note/ - Hermes Agentを育てる前に、サポートトリアージの合成シナリオを作る https://llm-lab.dev/posts/hermes-agent-002-support-triage-scenarios/ https://llm-lab.dev/en/posts/hermes-agent-002-support-triage-scenarios/ - Sakana Fugu APIをLangfuseで観測し、マルチエージェントの見えないコストを考える https://llm-lab.dev/posts/sakana-fugu-langfuse-experiment/ https://llm-lab.dev/en/posts/sakana-fugu-langfuse-experiment/ - Sakana Fuguを契約したので、マルチエージェントAPIの見えなさをLangfuseで観測する https://llm-lab.dev/posts/sakana-fugu-langfuse-observability-plan/ https://llm-lab.dev/en/posts/sakana-fugu-langfuse-observability-plan/ ## Series - Langfuse 朝ブリーフィング (Langfuse Morning Briefing): Langfuse observability practices - Vercel Eve 入門 (Getting Started with Vercel Eve): Workflow and automation with Eve - Flue 実践入門 (Flue Practical Guide): CI/CD pipeline experiments ## Markdown Export All articles are available in Markdown format for agent consumption: - `/posts/{slug}.md` — Japanese version - `/en/posts/{slug}.md` — English version ## License Unless otherwise stated, all content is provided for reference purposes. Code snippets are available under the MIT License unless noted. ## Contact Author: DUOps (デュオプス) X (Twitter): https://x.com/Crypto_LenLen ## Updated 2026-07-10T09:43:20.608Z