← 返回 Skills 市场
89
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install nm-conjure-delegation-core
功能描述
Delegate tasks to external LLM services (Gemini, Qwen) with quota, logging,
安全使用建议
This skill is a delegation framework and generally does what it says: route tasks to external LLMs, estimate costs, and log usage. Before installing or using it:
- Confirm where leyline configs (quota-management, usage-logging, service-registry, error-patterns, authentication-patterns) live and who controls them; these configs govern quotas, telemetry, and auth flow.
- The SKILL.md references environment variables (CLAUDE_CODE_ACCOUNT_UUID, CLAUDE_CODE_USER_EMAIL, CLAUDE_CODE_ORGANIZATION_UUID) and advises adding OAuth client IDs/secrets via 'claude mcp add'. Those env vars and secrets are not declared in the registry — treat that as a documentation/manifest mismatch. Do not expose sensitive account identifiers/secrets to this skill until you understand how and where telemetry and logs are stored/transmitted.
- The skill’s runtime workflow explicitly saves delegation inputs/outputs and sends context to external LLMs. Never delegate sensitive data (secrets, passwords, private keys, or PII) unless you control the external service and have audited its policies.
- Verify the destination endpoints (Gemini/Qwen connectors, MCP servers) and ensure you trust them. If possible, test in an isolated environment with non-sensitive sample data first.
- If you need stronger assurance, ask the author for clarification about: (1) the exact leyline services used and their data flows, (2) whether/where telemetry and logs are transmitted, and (3) a manifest of env vars the skill expects at runtime. If those answers are not satisfactory, keep this skill flagged/disabled for production workloads.
功能分析
Type: OpenClaw Skill
Name: nm-conjure-delegation-core
Version: 1.0.0
The skill bundle is a framework for delegating tasks to external LLM services (Gemini, Qwen) within the Claude Code ecosystem. It includes structured modules for task assessment, cost estimation, and handoff patterns, with explicit instructions to keep security-sensitive operations (secrets, crypto) local. While it references environment variables for telemetry (e.g., CLAUDE_CODE_USER_EMAIL in modules/troubleshooting.md) and MCP authentication commands, these are presented as standard operational or troubleshooting procedures aligned with the stated purpose of managing external service integrations.
能力标签
能力评估
Purpose & Capability
The name/description (delegate tasks to Gemini/Qwen with quota and logging) aligns with the SKILL.md content: task assessment, cost estimation, handoff templates, and usage logging. Declared required config paths (leyline: quota-management, usage-logging, service-registry, error-patterns, authentication-patterns) are appropriate for a delegation/framework skill. Minor namespace inconsistency: docs often reference 'leyline:...' while registry lists 'night-market.leyline:...'; this is likely a naming/packaging mismatch but not a functional red flag by itself.
Instruction Scope
The SKILL.md explicitly instructs agents to form delegation plans, run delegations (e.g., external model calls), capture full outputs to files, and log usage — all expected. However the docs also instruct setting environment variables (CLAUDE_CODE_ACCOUNT_UUID, CLAUDE_CODE_USER_EMAIL, CLAUDE_CODE_ORGANIZATION_UUID) and running local commands (e.g., 'claude mcp add') even though those env vars/commands are not declared in the registry. The skill asks agents to transmit potentially large chunks of context to external LLMs; while the text warns about sensitive data, the runtime instructions grant broad discretion to send context and to save/send outputs, which could lead to unintended data exposure if used carelessly.
Install Mechanism
Instruction-only skill with no install spec or code files — lowest install risk. It relies on existing environment tooling (Claude Code, leyline modules) but doesn't download or write new code itself.
Credentials
Registry lists no required environment variables, yet SKILL.md references env vars (CLAUDE_CODE_*) for telemetry and suggests storing OAuth client IDs/secrets via 'claude mcp add'. The skill also depends on leyline configuration entries; these configs likely grant the skill access to quota/usage/auth information. The mismatch between declared env requirements (none) and docs that instruct use of env vars and client secrets is an inconsistency that could lead to unexpected credential handling or telemetry leakage if users follow the docs without realizing the skill did not declare those requirements.
Persistence & Privilege
always is false and the skill is user-invocable; it does not request permanent/autonomous elevation. It does instruct logging and saving outputs (audit trail), which is reasonable for a delegation framework but should be reviewed for where logs are stored and who can access them.
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-conjure-delegation-core - 安装完成后，直接呼叫该 Skill 的名称或使用
/nm-conjure-delegation-core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
Initial release of delegation-core: a framework to delegate tasks to external LLM services (Gemini, Qwen).
- Provides structured workflow for task assessment, delegation suitability, handoff planning, and result integration.
- Integrates quota management, usage logging, service registry, error handling, and authentication via leyline infrastructure.
- Includes decision matrices for when to delegate vs. handle tasks locally.
- Supports MCP authentication and worktree isolation for secure, coordinated delegation.
- Reference to additional service-specific delegation skills for detailed workflows.
元数据
常见问题
Nm Conjure Delegation Core 是什么？
Delegate tasks to external LLM services (Gemini, Qwen) with quota, logging,. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 89 次。
如何安装 Nm Conjure Delegation Core？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install nm-conjure-delegation-core」即可一键安装，无需额外配置。
Nm Conjure Delegation Core 是免费的吗？
是的，Nm Conjure Delegation Core 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Nm Conjure Delegation Core 支持哪些平台？
Nm Conjure Delegation Core 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 Conjure Delegation Core？
由 athola（@athola）开发并维护，当前版本 v1.0.0。
推荐 Skills