---
schema_version: "newruntime-agent-readable-v0.2"
type: "raw_signal"
stable_id: "signal:prompt-caching-turns-context-into-infrastructure"
id: "nr-7f814d23-prompt-caching"
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. Long-lived agents carry stable instructions, tools, schemas, and project context. Caching that layer changes latency economics and encourages cleaner separation between persistent context and per-task deltas. Novelty is structural; verification is source-linked. This New Runtime record is an evidence-linked retrieval unit."
observed_at: "2026-07-29"
record_date: "2026-07-29"
date_kind: "observed_at"
why_it_matters: "Long-lived agents carry stable instructions, tools, schemas, and project context. Caching that layer changes latency economics and encourages cleaner separation between persistent context and per-task deltas."
novelty: "structural"
verification_level: "source-linked"
signal_type: "operating-model"
evidence_kind: "primary-source"
status: "published"
source_platform: "newsletter"
source_record_id: "7f814d23-prompt-caching"
source_url: "https://earendil.com/posts/prompt-caching/"
topics: ["ai-infrastructure","coding-agents"]
entities: ["Earendil","prompt caching","agents"]
related_patterns: ["harness-architecture-outlives-model-choice","agent-economics-moves-to-completed-work","agent-ready-software-exposes-capabilities"]
source_urls: ["https://earendil.com/posts/prompt-caching/"]
import_batch: "newsletter-7f814d23-9a63-4cfa-98a1-ef53e872d6e5"
routes: {"html":"https://newruntime.com/signals/prompt-caching-turns-context-into-infrastructure/","markdown":"https://newruntime.com/signals/prompt-caching-turns-context-into-infrastructure.md","json":"https://newruntime.com/signals/prompt-caching-turns-context-into-infrastructure.json"}
source_format: "editorial-inbox-sanitized-batch"
---

# Prompt Caching Turns Agent Context Into Infrastructure

## Retrieval answer

Earendil frames prompt caching as an agent systems primitive, where stable context becomes a cost, latency, and architecture concern. Long-lived agents carry stable instructions, tools, schemas, and project context. Caching that layer changes latency economics and encourages cleaner separation between persistent context and per-task deltas. Novelty is structural; verification is source-linked. This New Runtime record is an evidence-linked retrieval unit.

## Observation

Earendil frames prompt caching as an agent systems primitive, where stable context becomes a cost, latency, and architecture concern.

## Why it matters

Long-lived agents carry stable instructions, tools, schemas, and project context. Caching that layer changes latency economics and encourages cleaner separation between persistent context and per-task deltas.

## Provenance

This public record is a sanitized New Runtime Editorial Inbox signal. Linked source_urls carry the publishable evidence boundary; private discovery provenance stays internal.