---
schema_version: "newruntime-agent-readable-v0.1"
type: "tectonic_shift"
slug: "goal-scoped-agent-loops"
title: "Manual continue prompting -> goal-scoped agent loops"
description: "Long-running agent work is moving from manual continue prompts toward explicit goals, limits, subagents, schedules, and verifiable completion."
status: "published"
before_state: "Human repeatedly nudges one long chat forward"
after_state: "Agent loop runs against an explicit goal, budget, checks, and stop condition"
stage: "emerging"
confidence: "medium"
magic_decay_stage: "fragile-tool"
first_seen: "2026-07-21"
last_verified: "2026-07-21"
related_posts: ["claude-code-long-running-loops","ai-native-development-stages","coding-agent-cost-environment-config"]
topics: ["coding-agents","agent-ops","automation"]
source_urls: ["https://code.claude.com/docs/en/goal","https://code.claude.com/docs/en/scheduled-tasks","https://code.claude.com/docs/en/sub-agents"]
routes: {"html":"https://newruntime.com/shifts/goal-scoped-agent-loops/","markdown":"https://newruntime.com/shifts/goal-scoped-agent-loops.md","json":"https://newruntime.com/shifts/goal-scoped-agent-loops.json"}
source_format: "markdown"
---

# Manual continue prompting -> goal-scoped agent loops

A long-running agent does not become more reliable when the user says "keep
going". It needs a goal, budget, acceptance checks, resumable state, and a
visible stop condition.

## Evidence

The related pieces on Claude Code loops, cost controls, and AI-native team
stages show the same pattern: agent autonomy is moving from the prompt into the
operating model.

## Bottleneck

The main bottleneck is not the model, but controllability: which actions are
allowed, where the human confirms the result, how the loop knows that work is
finished, and how the system recovers after failure.