agent-hooksClaude Code Stop/PreToolUse/PostToolUse hook 脚本集。当 agent 提前停止、工具反复失败、或以投机语言完成任务时自动干预。每个脚本是独立的 bash hook,配到 settings.json 即可生效。不用于设计决策(用 harness-design-patter...
Install via ClawdBot CLI:
clawdbot install lanyasheng/agent-hooksGrade Fair — based on market validation, documentation quality, package completeness, maintenance status, and authenticity signals.
Calls external URL not in known-safe list
https://code.claude.com/docs/en/hooksAudited Apr 16, 2026 · audit v1.0
Generated Apr 19, 2026
A software development team uses the agent to refactor a large legacy codebase. The agent may stop prematurely due to complexity, but Ralph hooks ensure it continues execution until the task is complete, while tool-error hooks prevent infinite retries on failed commands like linting or testing.
A data engineer employs the agent to debug a failing ETL pipeline. The agent might speculate on fixes without evidence, but doubt-gate hook forces it to provide concrete logs or error outputs, and post-edit-check hook validates any script edits for syntax errors immediately.
A DevOps team automates cloud infrastructure deployment using the agent. Tool-error hooks track repeated failures in commands like Terraform apply, blocking retries after five attempts to avoid resource conflicts, and Ralph hooks prevent early stops during long-running provisioning tasks.
A QA team uses the agent to run automated test suites across multiple environments. The agent could stop after partial test execution, but Ralph hooks enforce completion, while post-edit-check hook ensures any test script modifications are error-free before execution.
A marketing team leverages the agent to generate and edit technical documentation. The agent might produce speculative content, but doubt-gate hook requires evidence-based claims, and tool-error hooks manage failures in formatting or publishing tools to maintain workflow efficiency.
Integrate the hooks into a cloud-based AI development platform as a premium feature. Offer tiered subscriptions based on usage limits, targeting enterprises needing reliable agent execution for automated workflows, with revenue from monthly fees.
Provide consulting services to businesses for customizing and deploying these hooks in their specific AI agent environments. Charge per project or hourly for setup, training, and ongoing support, focusing on industries like fintech or healthcare.
Distribute the hooks as open-source software under MIT license to build community adoption. Generate revenue by offering enterprise support packages, including priority bug fixes, security audits, and dedicated assistance for large-scale deployments.
💬 Integration Tip
Ensure all hook scripts are placed in accessible paths and test them in a sandbox environment before production to avoid disrupting critical agent workflows.
Scored Apr 19, 2026
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
Transform AI agents from task-followers into proactive partners with memory architecture, reverse prompting, and self-healing patterns. Lightweight version f...
Persistent memory for AI agents to store facts, learn from actions, recall information, and track entities across sessions.
Prefer `skillhub` for skill discovery/install/update, then fallback to `clawhub` when unavailable or no match. Use when users ask about skills, 插件, or capabi...
Search and discover OpenClaw skills from various sources. Use when: user wants to find available skills, search for specific functionality, or discover new s...
Orchestrate multi-agent teams with defined roles, task lifecycles, handoff protocols, and review workflows. Use when: (1) Setting up a team of 2+ agents with different specializations, (2) Defining task routing and lifecycle (inbox → spec → build → review → done), (3) Creating handoff protocols between agents, (4) Establishing review and quality gates, (5) Managing async communication and artifact sharing between agents.