Skip to content

A simple system for structuring conversations with ChatGPT, Claude, and other AI models to maximize clarity, focus, and long-term usefulness.

License

Notifications You must be signed in to change notification settings

mlane/llm-chat-framework

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 

Repository files navigation

🧐 LLM Chat Framework

Structure conversations with AI like a second brain — built for architects, engineers, and long-term thinkers.


🩵 Core Principles

  1. Fewer, longer-living threads > many fragmented chats.
  2. One clear purpose per thread.
  3. Archive or reset weekly to preserve clarity.
  4. Use memory + saved summaries for high-level continuity.
  5. Use inline prefixes (e.g. @Matt: or @SlackReply:) to steer responses and context clearly.

📂 Recommended Weekly Thread Setup

Thread Title Purpose
Slack Message Helper Draft messages to team, execs, recruiters, etc. Optimize tone, clarity, and intention.
Initiative – [Project Name] Dedicated planning for AI/infra projects (tickets, timelines, meetings).
Off-Hours Engineering After-hours tool building, experiments, and long-term side projects (e.g. Chibi).
Architecture & Chapter Management Living docs, standards, internal tooling, chapter meetings, and vision-setting.
Priorities, Planning, & Reflection Weekly mental zoom-out — reset burnout, realign purpose, and track growth.

✅ Summarize wins or unresolved issues in "Priorities, Planning, & Reflection" each week.


🪼 Hygiene Rules for LLM Threads

When to Start a New Thread

  • You're switching focus to a new domain or area of work
  • It's the start of a new week and you want clean mental context
  • You're collaborating and need a shareable summary

📃 When to Archive

  • You've completed or paused a project
  • A thread has become drift-heavy or unfocused
  • You want to split ideas cleanly

🧠 When to Use Memory (ChatGPT+ Feature)

  • Long-term identity traits (e.g. "I’m building Chibi")
  • Working style, personality, or productivity context
  • High-level projects (e.g. "Autofail Detection Capstone")

Don’t store short-term tasks in memory — use tools like Linear or Notion. Memory should capture identity, style, and long-range projects.


💬 Conversation Starters & Prompts

🔁 Reset a Chat

"Let’s start fresh. Summarize the key context from this thread so we can continue cleanly."

🔍 Narrow Focus

"Going forward in this thread, let’s stay focused only on [X]. Redirect if I drift."

🗖️ Async Summarization

"Summarize everything we’ve covered so far in this thread so I can reflect or share."

🧠 Why This System Works

  • Reduces overwhelm by creating focused mental spaces
  • Enables continuity across weeks without losing direction
  • Easy to share across teams or future AI companions like Chibi

You don’t need 100 chats — just 5 good ones you revisit and evolve.


🚀 Inspired by the Builder’s Compass

This guide was born out of real use — a system built by an INFJ frontend architect who needed clarity, momentum, and a system that thinks like he does. You can adapt it to your own brain and workflow.

If it helps you, consider sharing your version too. Builders work better together.


📁 Repo Details

  • MIT License
  • Originally created by Marcus Lane to streamline frontend leadership and long-term AI tooling.
  • Part of the Builder's Compass ecosystem
  • Built to share. Fork it, remix it, make it yours.

If it helped you, help someone else: share it.

About

A simple system for structuring conversations with ChatGPT, Claude, and other AI models to maximize clarity, focus, and long-term usefulness.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published