OReilly/Event/25/09/AI Codecon/1145 Context Engineering Addy Osmani

• Context Engineering – Person/Addy Osmani

  • Overview

    • Duration: 15 minutes
    • Time: 11:45
    • Speaker: Person/Addy Osmani
    • Event: OReilly/Event/25/09/AI Codecon
    • Resources: PDF

  • Description

    • Exploration of context engineering as "one small piece of an emerging thick layer of non-trivial software" that powers real LLM apps
    • Evolution of prompt engineering into a broader, more system-level approach
    • Practical, systems-based approach to structuring all the information an LLM needs to perform reliably and usefully

  • Key Topics

    • Dynamic assembly of task-specific context using data, instructions, tools, and memory
    • How good input design translates to better, more trustworthy output
    • Systems-level approach to LLM application development
    • Moving beyond simple prompt engineering

  • My Notes

    • "Context mismanagement is often a root cause of problems"
    • #Definition of Context Engineering
      • CONTEXT ENGINEERING

        MEANS PROVIDING AN AI WITH ALL THE
        INFORMATION AND TOOLS IT NEEDS TO
        SUCCESSFULLY COMPLETE A TASK – NOT JUST A
        CLEVERLY WORDED PROMPT.
        

    • 11:51 he has an example of IDE/Cline with a progress bar.
    • 11:51 he has a slide from Prompting 101 | Code w/ Claude - YouTube by Anthropic
    • 11:55 he mentions he uses IDE/Cline quite a bit, and he talks about 50% of context window as a threshold
    • 11:58 even more examples from IDE/Cline of AI/Context/Compact
    • mentions LangChain - Context Engineering
    • 11:59 he mentions How to Fix Your Context by Person/Drew Breunig
      • 
      • TODO follow him on Readwise
    • TODO follow Person/Addy Osmani in readwise