Atlasof World Models
Start

Courses · 6 pieces

Context engineering

6 mission decks on what a model sees and when. Prompt caching, retrieval, memory, tool design and agents that run for hours.

Decks

  1. mission-cards · 63 min · 16 cards

    What context engineering is

    The window as a budget, what goes in, order, and why it beats prompt tweaking

  2. mission-cards · 73 min · 19 cards

    Prompt caching

    cache_control, prefix order, TTL, minimum lengths, batch caching and your call sites

  3. mission-cards · 69 min · 18 cards

    Retrieval and grounding

    Graph modes, RRF, citations, and what to retrieve versus keep resident

  4. mission-cards · 72 min · 19 cards

    Memory

    memdb episodes and lessons, the MEMORY.md budget, harness memory writer and compaction

  5. mission-cards · 72 min · 19 cards

    Progressive disclosure and tool design

    Skills on demand, the tool catalogue budget, structured observations and sub-agents

  6. mission-cards · 79 min · 20 cards

    Long-running agents

    Checkpoints, evidence files, handoffs, the Sessions hub and context rot

Search the atlas