I Cut 80%+ of Context Overhead in My Coding Agent
asm3r96
12 points
3 comments
August 28, 2026
Related Discussions
Found 5 related stories in 68.8ms across 4,827 title embeddings via pgvector HNSW
- Code mode yields a 99.2% cost reduction in our systems tarasyarema · 61 pts · July 23, 2026 · 54% similar
- Agentic Context Management: Memory and Cost as Architecture Problems gdad · 17 pts · August 26, 2026 · 52% similar
- OpenAI reduces Codex Model Context Size from 372k to 272k AmazingTurtle · 336 pts · July 19, 2026 · 52% similar
- Agent-talk: Enabling coding agents to work together xhluca · 48 pts · July 16, 2026 · 51% similar
- Six months of writing code exclusively with agents bryanmikaelian · 66 pts · August 27, 2026 · 51% similar
Discussion Highlights (2 comments)
mnahkies
Does anyone know what the impact on prompt caching is? I like the idea of dynamic tool/skill sets, but wonder if it needs to be part of the new session workflow where you would choose a group to activate rather than this dynamic.
k9294
I'm pretty sure it's a bad idea because each time you connect and disconnect tools, you are hitting a full cache miss on the full context, that will be probably more expensive than having these tools in place from the very beginning.