---
schema_version: "newruntime-agent-readable-v0.1"
type: "trend_pattern"
slug: "goal-scoped-loops-replace-manual-continuation"
title: "Goal-scoped loops replace manual continuation"
description: "Long-running agent work is moving from repeated continue prompts toward explicit goals, stopping conditions, schedules, task state, and evaluator-driven retries."
thesis: "An autonomous loop becomes operationally useful when completion is an external contract rather than a model's subjective decision to stop."
status: "published"
confidence: "high"
first_seen: "2026-05-14"
last_verified: "2026-07-07"
supporting_signals: ["tg-2338","tg-2434","tg-2480","tg-2554","tg-2622","tg-2628"]
related_posts: ["claude-code-long-running-loops","linear-loops-agent-workflows","sierra-horizon-long-running-agent"]
counter_evidence: ["Open-ended research and product judgment often lack a machine-checkable completion condition and still require interactive steering.","A measurable target can be gamed or optimized at the expense of unmeasured quality, safety, and maintainability."]
revision_trigger: "Revise the thesis if unsupervised continuation prompts match goal contracts on completion quality, budget control, recovery, auditability, and operator effort."
topics: ["agent-loops","goals","automation"]
source_urls: ["https://code.claude.com/docs/en/goal","https://github.com/steipete/agent-scripts","https://cursor.com/changelog/06-18-26","https://github.blog/changelog/2026-06-17-github-copilot-app-generally-available"]
routes: {"html":"https://newruntime.com/patterns/goal-scoped-loops-replace-manual-continuation/","markdown":"https://newruntime.com/patterns/goal-scoped-loops-replace-manual-continuation.md","json":"https://newruntime.com/patterns/goal-scoped-loops-replace-manual-continuation.json"}
source_format: "markdown"
---

# Goal-scoped loops replace manual continuation

## What replaces keep going

Repeated continuation leaves the stopping rule inside the same model that is
doing the work. Goal-scoped systems move it into observable state: an
acceptance test, score, deadline, retry budget, queue state, or approval. The
agent can then continue, recover, or stop for a reason the operator can inspect.

## Supporting evidence

- Claude Code goals bind work to an explicit outcome.
- Maintainer scripts package recurring repository work as stateful loops.
- Cursor and GitHub turn successful interactive workflows into automations.
- Continuous evaluation gives multi-agent systems an external quality signal.

## Operational consequence

Every loop should declare its target, evaluator, budget, permissions, retry
policy, checkpoint, and escalation path. Tasks without a credible evaluator
should remain interactive rather than receiving artificial autonomy.