How MCP authorization works and what it leaves to youHow MCP authorization works under OAuth 2.1, why the agent should never hold the key, and where to check, mint and audit each tool call.27 min read·bigcircle.hashnode.dev6h ago
Human in the loop AI agents and when a person must approveWhich AI agent writes need a person to approve them, what the approver should see, where to enforce it, and how to retire approval one action at a time.17 min read·bigcircle.hashnode.dev2d ago
How LLM inference works and how to optimize itLLM inference runs in two phases: prefill reads the prompt, decode writes the answer. What to measure, cache and change for faster, cheaper inference.13 min read·bigcircle.hashnode.dev4d ago
How to set AI agent permissions and access controlSet AI agent permissions one action at a time: give each change its own tool, check authorization in code, and keep credentials away from the model.16 min read·bigcircle.hashnode.dev6d ago
Why RAG systems fail in productionRAG in production usually fails quietly: the answer sounds right, cites a nearby passage, and misses the fact the user needed.33 min read·bigcircle.hashnode.dev1ANov 8