Four diagnostic branches from the article: request rejection, a length stop, context shifting, and slow repeated prefill without an overflow. The figure summarizes source analysis, not new measurements.

llama-server Context Overflow: Diagnose Prompt Limit Failures

An agent that has been working against a self-hosted llama.cpp server for twenty or forty turns eventually stops working, and the symptom is almost never specific enough to act on: a half-written answer, an empty completion, a tool call that never arrives, or a request that never gets a reply at all. The tempting response is to raise -c, which is exactly the fix that KV memory may not allow, and which does nothing for the two of the four failure modes below that are not really about running out of room. ...

October 30, 2026 · 15 min · baeseokjae

MCP stdio Leaks: Why Your Agent's Tool Servers Hang at Shutdown, Die on a Bad Filename and Leave Ghost Tools

A stdio MCP server that never exits is not a mystery: the Python SDK’s writer task can be blocked in a pipe write that only the reader can unblock, so cancelling it waits forever. In MCP Python SDK 2.2.0 on Linux, this probe reproduced that hang and three related lifecycle defects — an unencodable string that kills the connection with no JSON-RPC error, a rejected ClientSessionGroup registration that leaves a second server child alive and unreachable, and a disconnected session whose old tool name survives in group.tools. All four have a host-side workaround you can enforce today, and the probe below counts exactly what your host leaks. ...

October 30, 2026 · 8 min · baeseokjae
The read/write asymmetry described for self-hosted mem0ai 2.2.1: an expired row can be hidden by reads yet remain available to the write-path duplicate check. The hosted platform is excluded.

mem0 Expired Memories: Why New Writes Get Blocked

Your agent stored user prefers aisle seats with an expiration date. The date passed. The user says it again. You call add() with a fresh expiration_date, get no exception, get an empty results list, and get_all() shows nothing. The fact never lands, and it will never land as long as the expired copy is what the search returns. On mem0ai 2.2.1 — the Python OSS package, self-hosted — the cause is a read/write asymmetry. Expiration is applied as a filter on the read paths but not on the ADD path. The expired row is still handed to the extraction LLM as an “existing memory”, and its hash is still loaded into the dedup set, so the fresh write is dropped as a duplicate of a record the user can no longer see. The same write machinery adds a second dependency: an update that only touches metadata or the expiry date still calls the embedding provider, and a provider error stops that update before anything is persisted. ...

October 28, 2026 · 13 min · baeseokjae
Agent Walking-Dead State Detection: How to Detect and Patch Stalled AI Agents

Agent Walking-Dead State Detection: How to Detect and Patch Stalled AI Agents

A “walking-dead” AI agent is a process that keeps running, consuming tokens and passing every liveness check, while producing no durable progress toward its task. You detect it not by watching whether the process is alive, but by tracking whether its work actually advances. In one documented case, a SWE-bench Verified agent stayed busy until it used 1.06 million tokens—inspecting, reasoning, and calling tools—yet never wrote the patch, and every health check reported it healthy. This guide shows you how to detect walking-dead states, separate liveness from progress, and patch stalled agents before they burn your budget. ...

September 23, 2026 · 11 min · baeseokjae
MCP Snoop Review 2026: Wireshark for MCP — Transparent Proxy for Tool Call Debugging

MCP Snoop Review 2026: Wireshark for MCP — Transparent Proxy for Tool Call Debugging

What is mcpsnoop? — Wireshark for MCP mcpsnoop is an open-source transparent proxy for the Model Context Protocol (MCP) that sits between your MCP client and server, capturing every JSON-RPC frame in real time. Built in Go with a Bubble Tea terminal UI, it gives developers the same kind of deep packet inspection for MCP traffic that Wireshark provides for network packets. Launched in late June 2026, mcpsnoop reached 270 GitHub stars in just three weeks and is rapidly becoming the go-to tool for MCP debugging. ...

July 16, 2026 · 11 min · baeseokjae
Coding Agent Debug Logs Guide 2026: Claude Code, Codex, GitHub MCP, and Playwright MCP

Coding Agent Debug Logs Guide 2026: Claude Code, Codex, GitHub MCP, and Playwright MCP

AI coding agents ship code faster than ever, but when they break, the debugging experience is nothing like a traditional stack trace. You don’t get a line number and a segfault — you get a silent hang, a garbled terminal, a 529 from the API, or an MCP server that just won’t connect. After spending the last year running Claude Code, Codex CLI, GitHub MCP Server, and Playwright MCP in production pipelines, I’ve collected the debug patterns that actually work. Here’s the field guide I wish I’d had. ...

July 4, 2026 · 11 min · baeseokjae