← 返回 Skills 市场
upstage-deployment

solar-delegation

作者 upstage-deployment · GitHub ↗ · v1.0.0
cross-platform ⚠ suspicious
270
总下载
1
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install solar-delegation
功能描述
Delegate longer user-facing text generation to Upstage Solar Pro3 while keeping the primary model for planning and tool calls. Use when drafting explanations...
安全使用建议
This skill appears to do what it says (delegate long-form generation to Upstage Solar via OpenRouter) but the package metadata fails to declare the API keys and config changes it needs. Before installing: (1) confirm you or an admin are willing to add an OpenRouter provider and store its API key in your gateway config (and understand where/how keys are stored), (2) verify the OpenRouter domain (https://openrouter.ai) is the intended endpoint, (3) ensure the skill will not be given broader credentials than needed (don’t reuse admin secrets unnecessarily), (4) ask the skill author/registry maintainer to update metadata to declare required env vars/credentials and document exactly what gateway config changes are required, and (5) consider limiting delegation to specific sessions or setting a conservative default threshold until you’ve tested behavior. If you cannot verify those points, treat the metadata omission as a red flag and do not enable this skill in production systems.
功能分析
Type: OpenClaw Skill Name: solar-delegation Version: 1.0.0 The skill bundle is benign. It describes a legitimate delegation mechanism for AI agents to offload long text generation to a specialized model (Upstage Solar Pro3 via OpenRouter). The `SKILL.md` instructions for the agent are clear, focused on its stated purpose, and notably include a critical safeguard: `중요: 도구(tool)를 사용하지 말고 텍스트만 바로 출력해줘. 파일 읽기/쓰기 등 도구 호출 금지.` (Important: Do not use tools, just output text directly. Prohibit tool calls like file read/write.), which actively mitigates prompt injection risks against the delegated model. The `references/setup-guide.md` provides standard configuration steps for OpenClaw. There is no evidence of data exfiltration, malicious execution, persistence mechanisms, or obfuscation.
能力评估
Purpose & Capability
The skill's stated purpose (route long user-facing text to Upstage Solar Pro3 while keeping the primary model for planning/tool calls) matches the runtime instructions: estimating token length, checking session enablement, and spawning a Solar Pro3 session via sessions_spawn. Requesting an OpenRouter provider and configuring Solar in the gateway is consistent with that purpose.
Instruction Scope
SKILL.md stays on task: estimate output tokens, check session enablement, spawn Solar, and forward results. It also instructs keeping orchestration minimal and to include a 'no tool call' instruction in spawn requests. However, the instructions explicitly require modifying persistent memory/config (storing thresholds and per-session enablement) and gateway provider config (adding OpenRouter + API key). Those are broader actions than simple runtime behavior and should be surfaced in metadata/permissions.
Install Mechanism
This is an instruction-only skill with no install spec or code files, so it does not write or execute new artifacts on disk by itself. That lowers installation risk; however, it directs manual edits to the gateway config and restarting the gateway, which are outside the skill bundle and require administrator action.
Credentials
The SKILL.md and setup guide require an OpenRouter API key (sk-or-...) and reference a primary model API key, but the registry metadata lists no required environment variables or primary credential. This mismatch is significant: the skill will not function without adding provider credentials to the gateway, and those credentials are not declared in the skill metadata for user review.
Persistence & Privilege
The skill does not request always:true or autonomous-only elevation, but its setup instructions call for persistent changes: adding a provider to gateway config, storing API keys in that config, and writing delegation policy to persistent memory. Those changes are legitimate for this feature but are system-wide and should be made knowingly and securely by an admin.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install solar-delegation
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /solar-delegation 触发
  4. 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 solar-delegation** - Enables automatic delegation of long-form text generation to Upstage Solar Pro3, triggered by session settings and output token threshold. - Provides configurable thresholds to control when delegation occurs (default: 200 tokens). - Supports selective delegation based on session (e.g., main, Slack, Telegram). - Includes clear guidelines and fallback for when Solar Pro3 is unavailable or times out. - Ensures Solar’s output is passed through directly without extra formatting or metadata. - Allows users to modify delegation settings (thresholds, enabled sessions) dynamically.
元数据
Slug solar-delegation
版本 1.0.0
许可证
累计安装 0
当前安装数 0
历史版本数 1
常见问题

solar-delegation 是什么?

Delegate longer user-facing text generation to Upstage Solar Pro3 while keeping the primary model for planning and tool calls. Use when drafting explanations... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 270 次。

如何安装 solar-delegation?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install solar-delegation」即可一键安装,无需额外配置。

solar-delegation 是免费的吗?

是的,solar-delegation 完全免费(开源免费),可自由下载、安装和使用。

solar-delegation 支持哪些平台?

solar-delegation 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 solar-delegation?

由 upstage-deployment(@upstage-deployment)开发并维护,当前版本 v1.0.0。

💬 留言讨论