v2.3.0 — Message Count & Internal Reasoning Cap

What's New in v2.3.0

Accurate Message Counts in UI

When filtering chat items (👤 User, 🤖 Agent, 🧠 Agent Reasoning, 🔒 Internal Reasoning), you will now see the explicit number of block/messages next to the total line count in the UI. For instance:
🤖 Agent Messages ··········· 10,168 (45 Msg)
This makes it incredibly easy to know exactly how many conversational turns you are currently including vs. filtering out, independent of their raw token length.

Internal Reasoning Cap

We've split reasoning into 🧠 Agent Reasoning and 🔒 Internal Reasoning, giving you discrete granular controls for each via the UI using the ◀/▶ keys!

---

Install / Update

🐧 Linux / 🍎 macOS:

curl -sO https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py && python3 codex-md.py

🪟 Windows — Command Prompt:

curl -sO https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py && python codex-md.py

🪟 Windows — PowerShell 7:

Invoke-WebRequest -Uri "https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py" -OutFile "codex-md.py"; python codex-md.py

v2.2.1 — Smoother Message Cap Steps

What's New in v2.2.1

Smoother message cap progression

The per-type message caps (👤 User, 🤖 Agent, 🧠 Reasoning) now step through values more gradually between 10 and 100, instead of jumping too coarsely.

New step sequence:
ALL → 5 → 10 → 20 → 30 → 40 → 50 → 60 → 70 → 80 → 90 → 100 → 150 → 200 → 300 → 500

---

Install / Update

🐧 Linux / 🍎 macOS:

curl -sO https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py && python3 codex-md.py

🪟 Windows — Command Prompt:

curl -sO https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py && python codex-md.py

🪟 Windows — PowerShell 7:

Invoke-WebRequest -Uri "https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py" -OutFile "codex-md.py"; python codex-md.py

v2.2.0 — Windows Compatibility

What's New in v2.2.0

Full Windows Support

The script now runs natively on Windows without any extra dependencies. Previously, import tty and import termios (Unix-only modules) caused an immediate crash on Windows.

Changes:

• Replaced Unix-only tty / termios imports with a conditional block — they are only loaded on Linux/macOS
• Added a Windows keyboard handler using the built-in msvcrt module (no pip install needed)
• os.system('clear') replaced with a cross-platform _clear_screen() that uses cls on Windows
• Clipboard copy already worked on Windows via clip.exe — no change needed there
• Zero change in functionality or UI — everything works identically across all platforms

---

Install / Update

🐧 Linux / 🍎 macOS:

curl -sO https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py && python3 codex-md.py

🪟 Windows — Command Prompt:

curl -sO https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py && python codex-md.py

🪟 Windows — PowerShell 7:

Invoke-WebRequest -Uri "https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py" -OutFile "codex-md.py"; python codex-md.py

v2.1.0 — Granular Message Filtering

What's New in v2.1.0

Granular Per-Type Message Filtering

You can now independently control the "Last N" limit for each chat section — separately for User messages, Agent responses, and Agent Reasoning blocks. Previously there was no way to trim just one type without affecting others.

New TUI controls (navigate with ↑↓ then use ◀▶):

• 👤 User Message Cap — keep only the last N user messages
• 🤖 Agent Message Cap — keep only the last N agent responses
• 🧠 Agent Reasoning Cap — keep only the last N reasoning blocks

Each cap defaults to ALL (no limit). Steps: 5 → 10 → 20 → 50 → 70 → 90 → ... → 500.
The terminal output cap remains unchanged and operates independently.

---

🚀 Install / Update

🐧 Linux / 🍎 macOS:

curl -sO https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py && python3 codex-md.py

🪟 Windows — Command Prompt:

curl -sO https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py && python codex-md.py

🪟 Windows — PowerShell 7:

Invoke-WebRequest -Uri "https://raw.githubusercontent.com/MeXenon/codex-session-export/main/codex-md.py" -OutFile "codex-md.py"; python codex-md.py

Codex Session Export v2.0 - The Ultimate Pipeline

✨ Major Features in v2

• Interactive TUI: Navigate the session filter visually with keyboard arrows.
• Instant Installation: Download and execute directly via our secure curl one-liner.
• Clean Chat Pipeline: Intelligently strips raw IDE context, active-files, and tab background variables injected by models.
• Dynamic Output Capping: Keep massive terminal payloads tight by clipping output blocks dynamically.
• Parallel Session Engine: Export multiple Codex sessions simultaneously into formatted code.