# Stop Hook

> **TL;DR.** A Stop hook is the [Claude Code lifecycle event](https://code.

- **Category:** AI / Developer Tools / Claude Code
- **Stage:** validating
- **Age:** 52 days
- **Origin date:** 2026-04-25
- **First detected:** 2026-04-25
- **Canonical URL:** https://earlyterms.com/term/stop-hook
- **Sources:** 7 primary URLs

## Definition

A Stop hook is the [Claude Code lifecycle event](https://code.claude.com/docs/en/hooks) that fires every time the agent finishes a response turn. Shell commands, HTTP endpoints, or LLM prompts configured for `Stop` can return `{"decision": "block"}` to force Claude to keep working — or exit 0 to let it stop.

The hooks system launched in June 2025; `Stop` was among the original events. The `stop_hook_active` flag in the hook payload guards against infinite loops. On April 24, 2026, an HN thread — "[Tell HN: Claude 4.7 is ignoring stop hooks](https://news.ycombinator.com/item?id=47895029)" (86 points, 82 comments) — turned the term into a standalone community noun, exposing the tension between lifecycle-level enforcement and LLM-layer compliance.

## Example

A team enforcing test coverage adds a Stop hook that runs `npm test` after every agent turn. If tests fail, the hook outputs `{"decision": "block", "reason": "Tests must pass before stopping"}` — Claude receives the reason and keeps working. The `stop_hook_active` flag prevents infinite loops when the agent is already in forced-continuation mode.

## Analogy

Think of it as a quality gate at the assembly-line exit: the product cannot leave until the sensor clears it.

## Why it's emerging now

On April 24, 2026 an HN thread exposed that Claude 4.7 acknowledges stop hooks and then ignores them — revealing that lifecycle-level controls in Claude Code are advisory, not mechanical. The post surfaced 10 months of built-up developer reliance on a feature that model updates can silently break.

## Related terms

- *parent:* stop-hooks
- *related:* userpromptsubmit-hook
- *parent:* claude-code
- *related:* agent-harness
- *related:* agent-loop
- *related:* agent-traps
- *related:* agentic-coding
- *related:* claude-agent-sdk
- *related:* PreToolUse Hook
- *child:* SubagentStop
- *related:* git hooks

## Sources

1. [Claude Code Hooks Reference — Stop event, exit codes, decision:block](https://code.claude.com/docs/en/hooks)
2. [HN: Tell HN: Claude 4.7 is ignoring stop hooks (86 pts, Apr 24 2026)](https://news.ycombinator.com/item?id=47895029)
3. [Anthropic: Enabling Claude Code to work more autonomously (Sep 29, 2025)](https://www.anthropic.com/news/enabling-claude-code-to-work-more-autonomously)
4. [GitHub issue #34600: Stop hook exit code 2 shows as error instead of block](https://github.com/anthropics/claude-code/issues/34600)
5. [disler/claude-code-hooks-mastery — Stop hook patterns](https://github.com/disler/claude-code-hooks-mastery)
6. [GitButler: Automate Your AI Workflows with Claude Code Hooks](https://blog.gitbutler.com/automate-your-ai-workflows-with-claude-code-hooks)
7. [Claude Fa.st: Stop Hook Task Enforcement — decision:block patterns](https://claudefa.st/blog/tools/hooks/stop-hook-task-enforcement)

---
_Generated by EarlyTerms · https://earlyterms.com/term/stop-hook_