ai-architecture-harness-zh
建立和使用 AI 编程架构护栏,防止架构坍缩、功能回退和长对话迭代漂移。Use when the user mentions AI 编程, Agent 编程, 架构坍缩, Harness Engineering, 设计意图, 验收规则, 黄金法则, 架构测试, 或希望让代码库更适合 AI Agent 安全修改。 Skill: ai-architecture-harness-zh Owner: hgvgfgvh Summary: 建立和使用 AI 编程架构护栏,防止架构坍缩、功能回退和长对话迭代漂移。Use when the user mentions AI 编程, Agent 编程, 架构坍缩, Harness Engineering, 设计意图, 验收规则, 黄金法则, 架构测试, 或希望让代码库更适合 AI Agent 安全修改。 Tags: latest:1.0.0 Version history: v1.0.0 | 2026-05-19T15:27:22.412Z | user Initial release of ai-architecture-harness-zh: - Introduces a four-layer guardrail model for AI programming architecture safety
Rank
62
Safety
84
Downloads
1.3k
Updated
Oct 10, 2026
Version
1.0.0
Source
CLAWHUB
About
What it does, and when to use it.
Capability contract not published. No trust telemetry is available yet. 1.3K downloads reported by the source. Last updated 10/10/2026.
Avoid when
- Contract metadata is missing or unavailable for deterministic execution.
Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing
Public facts
Every fact links back to the source it came from.
- Vendor
- Clawhubvendor · observed Oct 10, 2026
- Protocol compatibility
- OpenClawcompatibility · observed Oct 10, 2026
- Adoption signal
- 1.3K downloadsadoption · observed Oct 10, 2026
- Latest release
- 1.0.0release · observed May 19, 2026
- Handshake status
- UNKNOWNsecurity
Install and run
Setup complexity: low.
clawhub skill install s174x9jg32fvtnq226ytye9w3d83h6tm:ai-architecture-harness-zh- Setup complexity is LOW. This package is likely designed for quick installation with minimal external side-effects.
- Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.
Contract: missing
curl -s "https://www.xpersona.co/api/v1/agents/clawhub-hgvgfgvh-ai-architecture-harness-zh/snapshot"
Documentation
CLAWHUB
6,667 characters of source documentation, loaded on request.
Extracted files
3 files captured from the source.
SKILL.md
--- name: ai-architecture-harness-zh description: 建立和使用 AI 编程架构护栏,防止架构坍缩、功能回退和长对话迭代漂移。Use when the user mentions AI 编程, Agent 编程, 架构坍缩, Harness Engineering, 设计意图, 验收规则, 黄金法则, 架构测试, 或希望让代码库更适合 AI Agent 安全修改。 --- # AI 编程架构护栏 ## 目标 帮助编码 Agent 在大型或复杂项目中修改代码时,不破坏已有架构、核心功能和长期设计意图。 使用本 Skill 时,把代码仓库视为事实来源,但把人工维护的设计意图视为最高层指导。不要依赖长对话历史保存关键上下文,关键规则必须沉淀到仓库文档或可执行检查中。 ## 核心模型 使用四层护栏模型: ```text 1. 人工设计意图层 2. Agent 同步的架构文档和验收文档层 3. 硬性自动化约束层 4. 人工巡检和黄金法则反馈层 ``` Agent 的任务不是自由发挥,而是在明确边界和反馈回路中安全执行。 ## 推荐文档结构 为项目创建或完善护栏时,优先使用这个最小结构: ```text AGENTS.md docs/DESIGN_INTENT.md docs/ARCHITECTURE.md docs/ACCEPTANCE_RULES.md docs/GOLDEN_RULES.md docs/ARCHITECTURE_DRIFT.md ``` `DESIGN_INTENT.md` 由人工维护,记录项目目标、核心架构原则、不可破坏的取舍、历史设计决策。它是“宪法”,不要让 Agent 用当前实现覆盖人工意图。 `ARCHITECTURE.md` 记录当前确认过的架构地图。它可以由 Agent 基于代码和 `DESIGN_INTENT.md` 阶段性同步,但不能把偶然漂移自动合法化。 `ACCEPTANCE_RULES.md` 记录核心功能、架构承诺和非回归行为的验收方式。 `GOLDEN_RULES.md` 记录真实事故中沉淀出的强规则。每条规则应包含事故来源、禁止行为、正确做法和自动化检查方式。 `ARCHITECTURE_DRIFT.md` 记录设计意图和当前实现之间的差异,并分类为:符合意图、合理演进、技术债、需要人工决策、违反设计。 ## 开始编码前 在执行非平凡代码修改前: 1. 读取 `AGENTS.md`,如果存在。 2. 读取 `docs/DESIGN_INTENT.md`、`docs/ARCHITECTURE.md`、`docs/ACCEPTANCE_RULES.md`、`docs/GOLDEN_RULES.md`,如果存在。 3. 明确本次修改不能破坏的架构边界和行为。 4. 在编辑前说明本次修改范围。 5. 不要主动进行大范围重构、重命名、迁移或抽象改造,除非用户明确要求。 如果项目还没有这些护栏文档,先创建最小可用版本,不要一次性发明庞大的文档体系。 ## 设计意图维护流程 人工设计意图是最高层锚点。更新 `DESIGN_INTENT.md` 时: 1. 保留历史决策,不要简单覆盖旧内容。 2. 用日期或决策记录追加新意图。 3. 记录“为什么变化”,而不只是“变化成什么”。 4. 保持足够短,使 Agent 能在编码前读完。 5. 不允许 Agent 用实现便利性替代人工设计取舍。 推荐格式: ```markdown ## YYYY-MM-DD - [决策标题] 设计意图: [系统必须保持或演进成什么。] 原因: [为什么这个方向重要。] 影响: - [未来修改必须遵守的约束。] - [Agent 不允许破坏的内容。] ``` ## 架构同步流程 阶段性工作完成后,或重大任务开始前,同步架构文档: 1. 读取 `DESIGN_INTENT.md`。 2. 检查相关代码实现。 3. 对比设计意图和当前代码。 4. 对每个差异进行分类: - `符合意图` - `合理演进` - `技术债` - `需要人工决策` - `违反设计` 5. 只把确认过的架构写入 `ARCHITECTURE.md`。 6. 把未确认或可疑差异写入 `ARCHITECTURE_DRIFT.md`。 不要默认“当前代码就是正确架构”。当前代码可能已经发生坍缩或漂移。 ## 架构验收测试 把架构承诺转成可执行检查。对于每个架构基石: 1. 从 `ARCHITECTURE.md` 中识别架构基石节点。 2. 基于代码找出能证明该基石仍然有效的特殊调用链、数据流或不变量。 3. 编写聚焦的单元测试、集成测试、lint 规则或结构检查。 4. 把检查加入正常验证流程。 示例: ```text 架构基石: PlanAgent 必须具备多轮持续记忆。 可观测规律: 连续 10 轮对话后,必须触发记忆管理链路中的 retrieve、summarize/update、persist。 验收测试: 模拟 10 轮对话,断言预期的 MemoryManager 方法被调用,且结果进入持久化或上下文构建流程。 ``` 优先使用确定性检查,而不是只依赖 AI 判断: ```text 类型检查 > 单元测试 > 集成测试 > 架构测试 > lint > CI > AI review > prompt 提醒 ``` AI review 只能作为语义补充,不能作为核心护栏。 ## 硬约束优先目标 构建护栏时,优先防止这些问题: - 跨层调用或 import 违反架构方向 - 绕过核心 service、repository、memory manager、validator、provider - 为已有子系统创建竞争性重复实现 - 删除、弱化或绕开回归测试 - 未更新验收规则就改变外部行为 - 用直接数据访问替代稳定抽象 - 引入无边界复杂度、全局状态或隐藏副作用 如果项目有清晰分层,应该把允许的依赖方向写成测试或 lint 规则。 ## 黄金法则流程 当已有护栏没有挡住架构坍缩、功能回退或漂移时: 1. 总结事故。 2. 判断为什么现有文档或测试没有阻止它。 3. 在 `GOLDEN_RULES.md` 中新增规则。 4. 新增或提出一个确定性检查来执行该规则。 5. 把规则链接到测试、lint、CI 或 review checklist。 推荐格式: ```markdown ## [规则标题] 事故来源: [发生了什么,什么时候发生。] 禁止行为: [Agent 不允许再做什么。] 必须行为: [正确的架构行
_meta.json
{
"ownerId": "kn76tv8qxh5ap072zm27re50gh835yje",
"slug": "ai-architecture-harness-zh",
"version": "1.0.0",
"publishedAt": 1779204442412
}skill-card.md
## Description: 建立和使用 AI 编程架构护栏,防止架构坍缩、功能回退和长对话迭代漂移。 This skill is ready for commercial/non-commercial use. ## Publisher: [hgvgfgvh](https://clawhub.ai/user/hgvgfgvh) ### License/Terms of Use: MIT-0 ## Use Case: Developers and engineering agents use this Chinese-language skill to create, maintain, and apply architecture guardrails before modifying large or complex codebases. It guides documentation, acceptance checks, architecture drift review, and golden-rule feedback loops to reduce architecture collapse, regression, and long-conversation drift. ### Deployment Geography for Use: Global ## Known Risks and Mitigations: Risk: The skill may ask an agent to create or update repository documentation, tests, lint rules, or configuration. Mitigation: Review proposed repository changes like normal code changes before accepting them. Risk: Architecture guidance can become inaccurate if it treats current implementation drift as intended design. Mitigation: Compare updates against human-maintained design intent and keep uncertain differences in architecture drift documentation for review. ## Reference(s): - [ClawHub skill page](https://clawhub.ai/hgvgfgvh/skills/ai-architecture-harness-zh) ## Skill Output: **Output Type(s):** [Guidance, Markdown, Code, Shell commands, Configuration] **Output Format:** [Markdown guidance with optional repository documentation, tests, lint rules, shell commands, and configuration changes] **Output Parameters:** [1D] **Other Properties Related to Output:** [Chinese-language architecture guardrail guidance; repository changes should be reviewed like normal code changes.] ## Skill Version(s): 1.0.0 (source: server release metadata) ## Ethical Considerations: Users should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment.
AionUi
Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!
activepieces
AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents
cherry-studio
AI productivity studio with smart chat, autonomous agents, and 300+ assistants.
CopilotKit
The Frontend for Agents & Generative UI. React + Angular
Machine-readable data
The same record, as JSON, for agents and crawlers.
{
"facts": [
{
"factKey": "vendor",
"category": "vendor",
"label": "Vendor",
"value": "Clawhub",
"href": "https://clawhub.ai/hgvgfgvh/skills/ai-architecture-harness-zh",
"sourceUrl": "https://clawhub.ai/hgvgfgvh/skills/ai-architecture-harness-zh",
"sourceType": "profile",
"confidence": "medium",
"observedAt": "2026-10-10T17:32:23.119Z",
"isPublic": true
},
{
"factKey": "protocols",
"category": "compatibility",
"label": "Protocol compatibility",
"value": "OpenClaw",
"href": "https://www.xpersona.co/api/v1/agents/clawhub-hgvgfgvh-ai-architecture-harness-zh/contract",
"sourceUrl": "https://www.xpersona.co/api/v1/agents/clawhub-hgvgfgvh-ai-architecture-harness-zh/contract",
"sourceType": "contract",
"confidence": "medium",
"observedAt": "2026-10-10T17:32:23.119Z",
"isPublic": true
},
{
"factKey": "traction",
"category": "adoption",
"label": "Adoption signal",
"value": "1.3K downloads",
"href": "https://clawhub.ai/hgvgfgvh/ai-architecture-harness-zh",
"sourceUrl": "https://clawhub.ai/hgvgfgvh/ai-architecture-harness-zh",
"sourceType": "profile",
"confidence": "medium",
"observedAt": "2026-10-10T17:32:23.119Z",
"isPublic": true
},
{
"factKey": "latest_release",
"category": "release",
"label": "Latest release",
"value": "1.0.0",
"href": "https://clawhub.ai/hgvgfgvh/ai-architecture-harness-zh",
"sourceUrl": "https://clawhub.ai/hgvgfgvh/ai-architecture-harness-zh",
"sourceType": "release",
"confidence": "medium",
"observedAt": "2026-05-19T15:27:22.412Z",
"isPublic": true
},
{
"factKey": "handshake_status",
"category": "security",
"label": "Handshake status",
"value": "UNKNOWN",
"href": "https://www.xpersona.co/api/v1/agents/clawhub-hgvgfgvh-ai-architecture-harness-zh/trust",
"sourceUrl": "https://www.xpersona.co/api/v1/agents/clawhub-hgvgfgvh-ai-architecture-harness-zh/trust",
"sourceType": "trust",
"confidence": "medium",
"observedAt": null,
"isPublic": true
}
],
"events": [
{
"eventType": "release",
"title": "Release 1.0.0",
"description": "Initial release of ai-architecture-harness-zh: - Introduces a four-layer guardrail model for AI programming architecture safety. - Provides recommended documentation structure for maintaining design intent and architecture boundaries. - Details workflows for design intent updates, architecture synchronization, acceptance testing, and golden rule definition. - Emphasizes using documentation and automated checks over relying on long conversation history. - Outlines core risks and review steps to prevent architecture collapse, feature regression, and drift.",
"href": "https://clawhub.ai/hgvgfgvh/ai-architecture-harness-zh",
"sourceUrl": "https://clawhub.ai/hgvgfgvh/ai-architecture-harness-zh",
"sourceType": "release",
"confidence": "medium",
"observedAt": "2026-05-19T15:27:22.412Z",
"isPublic": true
}
]
}Record generated Oct 10, 2026.
