{"schema_version":"newruntime-agent-readable-v0.2","type":"post","stable_id":"post:prompt-caching-turns-context-into-infrastructure","slug":"prompt-caching-turns-context-into-infrastructure","title":"Prompt Caching Turns Agent Context Into Infrastructure","description":"Earendil frames prompt caching as an agent systems primitive, where stable context becomes a cost, latency, and architecture concern.","retrieval_nugget":"Earendil frames prompt caching as an agent systems primitive, where stable context becomes a cost, latency, and architecture concern. Earendil's prompt caching piece is important because it treats cache as architecture. Agent systems often carry a large stable prefix: instructions, tool schemas, policies, project conventions, and sometimes retrieved context.","status":"published","published_at":"2026-07-29","updated_at":"2026-07-29","record_date":"2026-07-29","date_kind":"published_at","topics":["ai-infrastructure","coding-agents"],"source_urls":["https://earendil.com/posts/prompt-caching/"],"visuals":[{"id":"prompt-caching-turns-context-into-infrastructure","kind":"editorial-diagram","role":"hero","src":"https://newruntime.com/images/posts/prompt-caching-turns-context-into-infrastructure.webp","alt":"Hand-drawn cache architecture diagram separating stable context, task delta, reused agent steps, cost, and harness feedback.","caption":"Prompt caching rewards a cleaner harness: stable context becomes reusable infrastructure while each task sends only the delta.","credit":"New Runtime synthesis","source_url":"https://earendil.com/posts/prompt-caching/","generated_with":"nano-banana-style-imagegen","width":1600,"height":900,"legend":[]}],"routes":{"html":"https://newruntime.com/posts/prompt-caching-turns-context-into-infrastructure/","markdown":"https://newruntime.com/posts/prompt-caching-turns-context-into-infrastructure.md","json":"https://newruntime.com/posts/prompt-caching-turns-context-into-infrastructure.json"},"source_format":"markdown","next_reads":[{"type":"topic","path":"/topics/ai-infrastructure/","reason":"Explore the ai infrastructure topic hub.","url":"https://newruntime.com/topics/ai-infrastructure/","title":"AI Infrastructure - New Runtime","media_type":"text/html"},{"type":"topic","path":"/topics/coding-agents/","reason":"Explore the coding agents topic hub.","url":"https://newruntime.com/topics/coding-agents/","title":"Coding agents - New Runtime","media_type":"text/html"},{"type":"related_material","path":"/posts/claude-md-prompt-cruft-shrinkage/","reason":"Shares ai infrastructure and coding agents.","url":"https://newruntime.com/posts/claude-md-prompt-cruft-shrinkage/","title":"Claude.md Shrinkage Says the Harness Is Learning What Not to Say","media_type":"text/html"},{"type":"related_material","path":"/posts/codex-hooks-close-the-type-error-loop/","reason":"Shares ai infrastructure and coding agents.","url":"https://newruntime.com/posts/codex-hooks-close-the-type-error-loop/","title":"Codex Hooks Close the Type Error Loop","media_type":"text/html"},{"type":"related_material","path":"/posts/codex-multi-agent-orchestration-skill/","reason":"Shares ai infrastructure and coding agents.","url":"https://newruntime.com/posts/codex-multi-agent-orchestration-skill/","title":"A Codex Skill Turns Multi-Agent Work Into a Reusable Control Surface","media_type":"text/html"}]}