Persistent conversational flow is a great approach, and it's what I'm using now that I've been using Codex extensively.

However, my personal approach is actually simpler and more direct.

Basically, I will open more than five dialog boxes with similar public capabilities and never close the window.

However, I need to assist some dialogue filtering mechanisms, because after a few days and dozens of rounds of context compression, the memory will become inaccurate.

Even if it is stored in the codex log, it will be inaccurate.

Moreover, according to my experience, the codex log will become extremely huge (I took the initiative to clean it).

I usually output important dialogue processes, conclusions and experiences into three different persistent files and put them in ob, so that people and codex can read them at any time.

The following short article is translated by Teacher Baoyu, and the article itself is shared by Codex official team members.

I have used several of the methods mentioned in it, but they all have some customized transformations, which will be shared in turn: ◦ An exclusive "chief of staff" dialogue flow (to help you handle daily chores) ◦ A dialogue flow specifically responsible for product release ◦ A dialogue flow responsible for reviewing documents ◦ A monitoring dialogue flow specifically focusing on external data They are actually your permanent workspace.

Codex can return to these conversations over time, remembering previous decisions you've made, your personal preferences, and your current progress.

Without this feature, you would have to refeed it this background information from scratch every time.

Visual summary

Article argument map

Generated from the post's content graph

Mermaid outline

flowchart LR
  format-long_post["longform note"]
  topic-memory["memory"]
  capability-product-surface["product surface"]
  format-long_post -->|covers| topic-memory
  format-long_post -->|signals| capability-product-surface

Visual structure

Essay structure map

Built from summary and key paragraph positions

Mermaid outline

flowchart LR
  thesis["Persistent conversational flow is a great approach, and it's what I'm using now that I've been using Codex..."]
  signal["Persistent conversational flow is a great approach, and it's what I'm using now that I've been using Codex..."]
  operator["Moreover, according to my experience, the codex log will become extremely huge (I took the initiative to cl..."]
  implication["Without this feature, you would have to refeed it this background information from scratch every time."]
  thesis -->|frames| signal
  signal -->|develops| operator
  operator -->|lands in| implication

Source: View the original post