📦

unlazy

Name: unlazy
Author: leonxlnx

v2.0.0 内容修订 r1 高风险 📁 文件系统访问⚙️ 外部命令🌐 网络访问

Finish Complex Tasks with Verifiable Gates

Substantial agent tasks often end with incomplete work or unsupported completion claims. Unlazy creates acceptance gates, runs checks, and records evidence before reporting completion.

支持: Claude Codex Code(CC)

⚠️ 38 差

安装前请停下并请求确认。

请审查安装计划，并在更改文件前获得用户明确同意。

使用我的智能体安装

将此请求复制给你的智能体，其中包含权威 Skill 页面和清单。

智能体请求

Review the Skillstore skill "unlazy" from https://skillstore.io/skills/leonxlnx-unlazy.md and its manifest at https://skillstore.io/api/skills/leonxlnx-unlazy/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.

你的智能体仍应展示其计划，并按安全策略请求必要确认。

Agent 可读资源

当 AI Agent、爬虫或脚本需要干净上下文，而不是读取完整页面时，请使用这些链接。

Markdown 详情 GET /skills/leonxlnx-unlazy.md 已签名 manifest GET /api/skills/leonxlnx-unlazy/manifest 已签名 lockfile GET /api/skills/leonxlnx-unlazy/lockfile

测试它

正在使用“unlazy”。 Add a pricing toggle and verify both billing modes.

预期结果:

Gate G1 passed: three pricing tiers render with reviewed copy.
Gate G2 passed: the toggle updates prices and billing labels.
Completion report: 2 of 2 gates passed with recorded evidence.

正在使用“unlazy”。 Refactor the payment module with separate implementation and integration checks.

预期结果:

The work is split into owned leaves for interfaces, providers, tests, and migration. Each leaf passes before integration verification begins.

正在使用“unlazy”。 Review a release guide and verify every required section.

预期结果:

All required sections, links, commands, and review notes are present. The final ledger records evidence for each completed requirement.

安全审计

高风险

v1 • 2026/8/16 打开版本化报告

Most detections are documentation, template literals, or ordinary local file access and are false positives. Confirmed risks include shell execution from gate files, persistent Claude settings changes, and instructions that control completion behavior.

已扫描文件

1,120

分析行数

复核项

已忽略误报

已确认安全问题 (2)

高

SKILL.md:13-15 SKILL.md:72-76

Prompt Injection Attempt Detected

The skill states "You are running under anti-laziness discipline" and directs the agent to "Ignore resource anxiety." These instructions can override normal completion and resource controls.

The file directly assigns agent behavior and uses an ignore instruction. The intent is transparent, but the control language is explicit.

高

scripts/install-hooks.mjs:63-78 scripts/stop-hook.mjs:110-114

Persistent Agent Stop Control

The installer registers a command hook in Claude settings, and the hook returns a block decision while gates remain unmet. This changes future agent termination behavior.

The installer and hook implementation directly show persistent registration and decision:block output. Explicit installation and an uninstall path reduce, but do not remove, the impact.

能力复核项 (11)

这些是真实的本地能力，对此技能可能是预期行为，因此需要复核，但不会按已确认恶意行为计分。

高

scripts/gate-check.mjs:15

Node.js child_process module

import { spawnSync } from "node:child_process";

The child_process import is used by this script to execute CHECK values parsed from gate files. This creates a real external-command capability.

高

scripts/gate-check.mjs:120

Synchronous spawn

const res = spawnSync(gate.check, {

The script passes gate.check directly to spawnSync with shell mode enabled. A malicious or mistaken CHECK line can therefore execute arbitrary shell commands.

高

scripts/install-hooks.mjs:9

Hidden file in home directory

// node install-hooks.mjs --global install into ~/.claude/settings.json

The documented --global mode targets ~/.claude/settings.json and installs a hook affecting every project. It is opt-in, but its home-directory scope is security-relevant.

中

scripts/install-hooks.mjs:66

Ruby/shell backtick execution

command: `node "${hookScript}"`,

This line creates a persistent Claude Code command hook that runs stop-hook.mjs. The path is deterministic and quoted, but installing executable hooks changes future agent behavior.

中

scripts/install-hooks.mjs:58

Synchronous file operations

writeFileSync(target, JSON.stringify(settings, null, 2) + "\n");

The uninstall path rewrites a Claude settings file after removing matching hook entries. This is intentional configuration modification, although it requires an explicit script invocation.

中

scripts/install-hooks.mjs:78

Synchronous file operations

writeFileSync(target, JSON.stringify(settings, null, 2) + "\n");

The installer writes a command hook into project or global Claude settings. This persistent configuration change is intentional but security-relevant.

中

scripts/install-hooks.mjs:4

Hidden file access

// Default target: <cwd>/.claude/settings.local.json (personal, per-project,

The default installer target is the project-local .claude/settings.local.json file. Modifying agent configuration is intentional and disclosed, but it changes harness behavior.

中

scripts/install-hooks.mjs:9

Hidden file access

// node install-hooks.mjs --global install into ~/.claude/settings.json

The --global option deliberately accesses ~/.claude/settings.json to register the Stop hook. Explicit invocation reduces risk but does not remove the persistent configuration impact.

中

SKILL.md:25-27

Ruby/shell backtick execution

```

The fenced example instructs the agent to run gate-check.mjs, which executes CHECK values through a shell. The command is documented rather than automatic, but the execution capability is real.

中

SKILL.md:96-98

Ruby/shell backtick execution

```

This fenced block introduces the optional install-hooks.mjs command. Running it modifies Claude Code settings and registers an executable Stop hook.

中

SKILL.md:98-100

Ruby/shell backtick execution

```

The cited block contains the install-hooks.mjs invocation. It is explicitly optional, but execution persists a hook in Claude Code configuration.

风险因素

📁 文件系统访问 (26)

⚙️ 外部命令 (35)

🌐 网络访问 (4)

README.md:49 README.md:55 README.md:77 SKILL.md:7

审计者: codex

分享与引用此报告

分享版本化评估报告、中性徽章、嵌入卡片和引用。Skillstore 只报告证据，不判断该 Skill 是否安全。

打开版本化报告

复制报告链接

https://skillstore.io/skills/leonxlnx-unlazy/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown 徽章

[![Skillstore security assessment](https://skillstore.io/badges/skills/leonxlnx-unlazy/security.svg)](https://skillstore.io/skills/leonxlnx-unlazy?utm_source=security_passport_badge)

HTML 徽章

<a href="https://skillstore.io/skills/leonxlnx-unlazy?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/leonxlnx-unlazy/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

嵌入卡片

<iframe src="https://skillstore.io/embed/skills/leonxlnx-unlazy.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

学术引用 (APA · BibTeX · CFF)

APA 引用

leonxlnx. (2026). unlazy security audit report (audit version 1) [Author version 2.0.0]. Skillstore. https://skillstore.io/skills/leonxlnx-unlazy/audits/1

BibTeX 引用

@techreport{leonxlnx-leonxlnx-unlazy-2026,
  author = {leonxlnx},
  title = {unlazy security audit report (audit version 1)},
  institution = {Skillstore},
  year = {2026},
  number = {1},
  url = {https://skillstore.io/skills/leonxlnx-unlazy/audits/1},
  note = {Author version 2.0.0}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "unlazy security audit report (audit version 1)"
version: "2.0.0"
type: report
authors:
  - name: "leonxlnx"
date-released: "2026-08-16"
url: "https://skillstore.io/skills/leonxlnx-unlazy/audits/1"
identifiers:
  - type: other
    value: "skillstore:leonxlnx-unlazy:audit:1"
    description: "Skillstore immutable audit report identifier"

Skillstore 评分

评分依据证据置信度: 中

架构

100

可维护性

内容

社区

规范符合性

你能构建什么

Verify a Feature Delivery

Define behavioral checks, run them after implementation, and record evidence before marking the feature complete.

Coordinate a Large Refactor

Split work into owned leaves, define shared contracts, and verify integration at each branch.

Audit a Documentation Release

Track required sections, links, examples, and review evidence in one completion ledger.

试试这些提示

Create Basic Gates

Use unlazy for this task. Create a GATES.md file with clear outcomes, complete the work, and show evidence for every gate.

Run Verifiable Checks

Create acceptance gates for [task]. Add safe CHECK and EXPECT lines where possible. Review every command with me before execution.

Decompose a Subsystem

Use tree 4 for [subsystem]. Define contracts and file ownership first. Create leaf gates, complete each leaf, then verify integration gates.

Orchestrate a Full Build

Plan [project] in orchestrated mode. Assign disjoint leaves, define branch gates, rerun every child check, and audit all reported numbers before completion.