OpenClaw / Octo AgentSkills for Mininglamp 工具
概览
一组用于使用 Mininglamp 工具的 OpenClaw / octo AgentSkills 合集。 每个子目录是一个自包含技能，带有一个 SKILL.md（以及可选引用/）。 最简单的方法是：它会帮你安装技能，然后启动入职程序。把这个复制粘贴到和你的机器人聊天里（替换技能名以安装另一个）： 机器人在底层操作（你不需要手动操作）： 复制后，机器人会读取新 SKILL.md 并运行——它会检查你的octo-multica密钥密钥是否已配置，如果没有，会引导你完成创建。所以从你这边来说，实际上就是：发送句子→回答密钥提示，→完成。 这些是标准的AgentSkills。在你的OpenClaw技能路径中放置一个技能目录（例如~/.openclaw/skills/），或者如果ClawHub发布了，则通过ClawHub安装。 然后代理会根据 SKILL.md 描述来识别。这里的大多数技能都有内置的首次入门流程，带你完成所需的密钥/密钥配置。
概述（AI 翻译）
一套用于使用 OpenClaw / 章鱼代理技能的合集 矿灯模具。
每个子目录是一个自包含技能，带有SKILL.md（以及可选的references/）。
技能
| 技巧 | 它的作用 |
|---|---|
octo-multica-cloud |
从octo/OpenClaw驱动云托管的Multica SaaS（multica.imocto.cn）——使用每个用户的octo-multica-key API密钥，读取/创建问题，列出代理，派遣任务，添加评论。账户无关;在运行时发现工作区和代理。 |
octo-loop ⚗️ experimental |
一个基于octoim + octo-spec + octo-multica-cloud的OpenClaw循环工程标准。一个三环嵌套合同（编排/执行/验证），具有循环不可作弊的接受门。通过角色占位符去品牌化;通过octo-multica-cloud实现账户无关。抢先体验——效果仍在试点数据中验证。 |
multica-squad-onboarding |
Multica小队的对话式冷启动顾问。 |
本页由 AI 翻译，技术细节请以原文为准。
README
以下为原文 README 节选，完整内容请查看原文。
octo-skills
A collection of OpenClaw / octo AgentSkills for working with Mininglamp tooling.
Each subdirectory is a self-contained skill with a SKILL.md (and optional references/).
Skills
| Skill | What it does |
|---|---|
octo-multica-cloud |
Drive the cloud-hosted multica SaaS (multica.imocto.cn) from octo/OpenClaw — read/create issues, list agents, dispatch tasks, add comments — using a per-user octo-multica-key API key. Account-agnostic; discovers workspaces and agents at runtime. |
octo-loop ⚗️ experimental |
An OpenClaw loop-engineering standard layered on octoim + octo-spec + octo-multica-cloud. A three-ring nested contract (orchestration / execution / verification) with a loop-uncheatable acceptance gate. De-branded with role placeholders; account-agnostic via octo-multica-cloud. Early access — efficacy still being validated by pilot data. |
multica-squad-onboarding |
Conversational cold-start advisor for Multica squads. Describe a goal + rough team and it clarifies the gaps over multi-turn conversation, designs a complete squad blueprint (agents with authored instructions, skills, leader, semantic roles), and — after approval — provisions and verifies it end-to-end via the multica CLI. Also handles incremental updates to existing squads. |
🚀 Quick install (one line to your octo bot)
The easiest way: paste one sentence to your OpenClaw/octo bot and it installs the skill for you, then runs onboarding. Copy-paste this into a chat with your bot (replace the skill name to install a different one):
帮我安装 octo-skills 里的 octo-multica-cloud 技能：从 https://github.com/Mininglamp-OSS/octo-skills 拉取，装到我的 OpenClaw skills 目录，然后跑首次引导帮我配置密钥。
English version:
Install the octo-multica-cloud skill from https://github.com/Mininglamp-OSS/octo-skills into my OpenClaw skills directory, then run its first-run onboarding to help me configure the key.
What the bot does under the hood (no manual steps for you):
# one-shot installer (clones repo, copies the skill into your skills dir)
curl -fsSL https://raw.githubusercontent.com/Mininglamp-OSS/octo-skills/main/install.sh | bash -s octo-multica-cloud
Or equivalently, by hand:
TMP=$(mktemp -d)
git clone --depth 1 https://github.com/Mininglamp-OSS/octo-skills.git "$TMP/octo-skills"
DEST="${OPENCLAW_SKILLS_DIR:-$HOME/.openclaw/skills}"
mkdir -p "$DEST"
cp -r "$TMP/octo-skills/octo-multica-cloud" "$DEST/"
rm -rf "$TMP"
echo "installed -> $DEST/octo-multica-cloud"
After copying, the bot reads the new SKILL.md and runs its first-run onboarding — it checks
whether your octo-multica-key secret is configured and, if not, walks you through creating it.
So from your side it really is just: send the sentence → answer the key prompt → done.
Manual install
推荐工具
换一个关键词，或者移除筛选条件。
安装
npx skillfish add mininglamp-oss/octo-skills