← 返回 Skills 市场
84
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install nm-leyline-risk-classification
功能描述
Inline risk classification for agent tasks using a 4-tier model. Hybrid routing: GREEN/YELLOW use heuristic file-pattern matching, RED/CRITICAL escalate to w...
安全使用建议
This is an instruction-only risk-classification skill and appears coherent with its stated purpose. Before installing, ensure you understand and permit the agent's access to: the repository (to read affected_files and run import/coverage analysis), your test runner (it may run related or full test suites for RED), and any project inbox or conflict-tracking system referenced. Check the night-market.error-patterns config (the skill declares this path) and verify the external skill attune:war-room-checkpoint (used for high-risk escalation) is trusted in your environment. No credentials are requested by this skill, but granting repository/test access is needed for it to work and increases the surface area — if you want a smaller blast radius, restrict the skill to read-only repo access or require human approval before escalation to war-room steps.
功能分析
Type: OpenClaw Skill
Name: nm-leyline-risk-classification
Version: 1.0.0
The skill bundle provides a comprehensive 4-tier risk classification framework (Green, Yellow, Red, Critical) for agent tasks. It uses heuristic file-pattern matching and readiness levels to determine necessary safety controls and verification gates, such as human approval for critical tasks. The logic is transparently documented across several files, including SKILL.md and heuristic-classifier.md, focusing on identifying high-risk operations like database migrations, security changes, and production deployments. No malicious code, data exfiltration, or prompt injection attempts were found.
能力标签
能力评估
Purpose & Capability
Name/description (risk classification, hybrid routing) match the SKILL.md contents: pattern-based heuristics for GREEN/YELLOW and escalation to Skill(attune:war-room-checkpoint) for RED/CRITICAL. Declared dependency on night-market.error-patterns is plausible for pattern config.
Instruction Scope
Instructions legitimately reference reading task.affected_files, matching file paths, running git status/conflict checks, and executing tests. They also mention checking a 'team inbox' for conflict alerts and performing dependency/import analysis (count_importers). Those actions require repository and project-system access; this is expected for a risk classifier but broadens the skill's runtime read-surface beyond just pattern matching.
Install Mechanism
Instruction-only skill with no install spec and no code files — lowest install risk. Nothing is downloaded or written to disk by the skill itself.
Credentials
No environment variables or credentials are required, which is proportionate. The only required config path is night-market.error-patterns; the SKILL.md also embeds pattern rules, so the explicit config requirement is plausible but not fully explained and should be checked.
Persistence & Privilege
always:false and no claims of modifying other skills or system-wide settings. The skill invokes other skills (attune:war-room-checkpoint) for escalation, which is expected; autonomous invocation of that skill is possible but normal for the platform.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install nm-leyline-risk-classification - 安装完成后，直接呼叫该 Skill 的名称或使用
/nm-leyline-risk-classification触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
- Introduces inline risk classification for agent tasks using a 4-tier (GREEN/YELLOW/RED/CRITICAL) model.
- Implements hybrid routing: GREEN/YELLOW risks are handled with heuristic file-pattern matching; RED/CRITICAL tasks are escalated to a war-room-checkpoint for reversibility scoring.
- Extends task metadata with an explicit risk tier and rationale for downstream consumption and automation.
- Clearly defines when to use the skill, the scope of each risk tier, and corresponding required verification controls.
- Includes module references and integration patterns for easy adoption in orchestrators and task generators.
元数据
常见问题
Nm Leyline Risk Classification 是什么？
Inline risk classification for agent tasks using a 4-tier model. Hybrid routing: GREEN/YELLOW use heuristic file-pattern matching, RED/CRITICAL escalate to w... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 84 次。
如何安装 Nm Leyline Risk Classification？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install nm-leyline-risk-classification」即可一键安装，无需额外配置。
Nm Leyline Risk Classification 是免费的吗？
是的，Nm Leyline Risk Classification 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Nm Leyline Risk Classification 支持哪些平台？
Nm Leyline Risk Classification 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Nm Leyline Risk Classification？
由 athola（@athola）开发并维护，当前版本 v1.0.0。
推荐 Skills