Working Notes on Agent Systems/Brad Zhang

@teach_fireworks / X longform

Why does Codex crash every time I open it? Finally found the reason! Because I am a heavy user of Codex,

Why does Codex crash every time I open it? Finally found the reason! Because I am a heavy user of Codex, I usually start with 6G of memory, and sometimes it will...

July 19, 2026 · 2 min read

Why does Codex crash every time I open it? Finally found the reason! Because I am a heavy user of Codex,
Figure 1 / source image

Why does Codex crash every time I open it? Finally found the reason! Because I am a heavy user of Codex, I usually start with 6G of memory, and sometimes it will increase several times, until it freezes. During this period, with Tibo's reset benefits, my macmini basically runs 7*24. The most effective, simple and direct way is to completely exit and restart Codex, hhh This is the advice Codex gave me: 1 Completely exit and reopen Codex, first clear the current 7 sets of residual MCP. Merely closing the window does not necessarily exit the background process. High-occupancy MCPs are turned off by default: 2 Keep the low-occupancy and commonly used node_repl, openai-developers, and cowart; if browser automation and image editing are rarely used, they can also be closed together. Limit the number of active tasks running at the same time to 1–2. Currently, there are only 2 active tasks, but 7 sets of tool processes have been accumulated. The more background tasks, the easier it is to amplify the problem. Optional Optimization Chronicle currently takes up approximately 0.29 GB. Turning off automatic memory/desktop recording can continue to reduce usage, but the historical context will be lost, so it is recommended to keep it for now. Avoid loading lots of images, video previews and browser pages continuously in the same long task, Renderer/GPU currently totals about 1.7 GB and may peak at over 3 GB. Completely exit the Codex every few days until MCP lifecycle issues are resolved in the new version. It is expected that after a full reboot and streamlining the high-occupancy MCP, normal can be reduced from about 5.7 GB to

  • 4–4.0 GB. The core App, Renderer, and GPU themselves still take up about 3 GB and are difficult to completely eliminate through configuration.

Visual summary

Article argument map

Generated from the post's content graph

FORMATTOPICCAPABILITYMARKETcoverscoverssignalssignalssignalssignalsFORMATlongform noteTOPICmemoryTOPICtoolingCAPABILITYagent workflowCAPABILITYAI-native workbenchCAPABILITYdeveloper toolingMARKETapplied AI team
Mermaid outline
flowchart LR
  format-long_post["longform note"]
  topic-memory["memory"]
  topic-tooling["tooling"]
  capability-agent-workflow["agent workflow"]
  capability-ai-native-workbench["AI-native workbench"]
  capability-developer-tooling["developer tooling"]
  market-applied-ai-team["applied AI team"]
  format-long_post -->|covers| topic-memory
  format-long_post -->|covers| topic-tooling
  format-long_post -->|signals| capability-agent-workflow
  format-long_post -->|signals| capability-ai-native-workbench
  format-long_post -->|signals| capability-developer-tooling
  format-long_post -->|signals| market-applied-ai-team

Visual structure

Essay structure map

Built from summary and key paragraph positions

Why does Codex crash every time I open it? Finally found the reason! Because I am a h...THESISWhy does Codex crashevery time I open it?Finally found thereason! Because I am aSIGNALWhy does Codex crashevery time I open it?Finally found thereason! Because I am aOPERATOR3.4–4.0 GB. The coreApp, Renderer, and GPUthemselves still takeup about 3 GB and areIMPLICATION3.4–4.0 GB. The coreApp, Renderer, and GPUthemselves still takeup about 3 GB and are
Mermaid outline
flowchart LR
  thesis["Why does Codex crash every time I open it? Finally found the reason! Because I am a heavy user of Codex, I..."]
  signal["Why does Codex crash every time I open it? Finally found the reason! Because I am a heavy user of Codex, I..."]
  operator["3.4–4.0 GB. The core App, Renderer, and GPU themselves still take up about 3 GB and are difficult to comple..."]
  implication["3.4–4.0 GB. The core App, Renderer, and GPU themselves still take up about 3 GB and are difficult to comple..."]
  thesis -->|frames| signal
  signal -->|develops| operator
  operator -->|lands in| implication

Source: View the original post