Agent 生成器
- 作者仓库星标 301
- 许可证 MIT
- 作者更新于 实时读取
- 作者仓库 marketplace
- 领域
- AI 智能
- 兼容 Agent
-
- Claude Code
- Cursor
- Cline
- Codex
- Windsurf
- Gemini CLI
- +20
- 信任分
- 98 / 100 · 已通过审计
- 作者 / 版本 / 许可
- @aiskillstore · v1.0.0 · MIT
- Token 消耗评级
- 低消耗
- 接入复杂程度
- 需简单配置
- 是否需要外部 API Key
- 不需要
- 兼容的系统
- macOS · Linux · Windows
- 底层运行要求
- 无特殊要求
- 文件与系统权限
-
- 只读
- 允许写入 / 修改
- Shell 执行
- 读取环境变量
- 网络行为
- 仅限本地
- 安装命令数
- 26 条
档案由构建时根据 SKILL.md 与安装命令自动衍生,可能与作者实际意图存在差异。
需要注意: 未限定 allowed-tools,默认拥有全部工具权限。
---
name: promote-skill
description: Use when publishing a SKILL.md-style agent skill across uGig, sh1pt, GitHub/gists, and follow-on…
category: AI 智能
runtime: 无特殊运行时
---
# promote-skill 输出预览
## PART A: 任务判断
- 适用问题:提示词、Agent 工作流、模型评估或自动化推理。
- 输入要求:目标材料、限制条件、期望输出和验收方式。
- 证据边界:围绕“Overview / When to Use / Inputs to collect or infer”读取原文规则,不把推断写成作者承诺。
## PART B: 执行结果
- **01** 任务判断:确认你的需求是否属于提示词、Agent 工作流、模型评估或自动化推理,并标出输入、限制和预期结果。
- **02** 执行计划:优先按“Overview / When to Use / Inputs to collect or infer”拆成步骤,说明每一步会读取什么、修改什么、产出什么。
- **03** 交付结果:给出可复制的命令、文件改动、检查清单或内容草稿,并说明如何继续迭代。
- **04** 风险边界:结合 读取文件、写入/修改文件、执行终端命令、读取环境变量、主要在本地完成、通常不需要额外 API Key 给出执行前确认项。
## Running Rules
- 读取文件、写入/修改文件、执行终端命令、读取环境变量;主要在本地完成;通常不需要额外 API Key。
- 先小样例验证,再放大到真实任务。
- 交付时同时给结果、检查口径和下一步迭代建议。 原文出现了 `/skills`、`/tmp` 这类斜杠命令;如果你的 Agent 支持命令触发,优先用命令开场,再补充目标和边界。
告诉 Agent 目标文件或材料、期望结果、不可改范围、是否允许联网或执行命令。本 Skill 的权限画像是:读取文件、写入/修改文件、执行终端命令、读取环境变量。
先用一个小任务确认它会围绕“Overview / When to Use / Inputs to collect or infer”工作;涉及文件或命令时,先看 diff、日志、预览或测试结果。
检查最终产物是否包含明确结果、必要证据和下一步动作;如果输出泛泛而谈,就补充输入、边界和验收标准后重跑。
---
name: promote-skill
description: Use when publishing a SKILL.md-style agent skill across uGig, sh1pt, GitHub/gists, and follow-on…
category: AI 智能
source: aiskillstore/marketplace
---
# promote-skill
## 什么时候使用
- 把 AI / Agent方向的常用动作沉淀成 Agent 可调用的技能 适合处理AI Agent、提示词、模型评估与自动化推理,核心价值是把输入、判断、执行、验证和交付边界固定下来,避免 Agent 泛泛回答。 把任务拆成可执行、可检查…
- 面向提示词、Agent 工作流、模型评估或自动化推理,优先处理能明确输入、步骤和验收标准的工作。
## 需要提供什么
- 目标材料、目录范围、期望结果和不可改动内容。
- 是否允许联网、执行命令、读写文件或调用外部服务。
## 执行规则
- 围绕「Overview / When to Use / Inputs to collect or infer」组织步骤,不把推断写成作者事实。
- 读取文件、写入/修改文件、执行终端命令、读取环境变量;主要在本地完成;通常不需要额外 API Key。
- 先跑小样例,确认结果可检查后再扩大任务范围。
## 输出要求
- 给出最终产物、关键证据、验证方式和下一步动作。
- 信息不足时标记 unknown,不编造命令、平台或依赖。 作者原文负责流程事实;仓库文件负责来源和命令;流狐只补充适用场景、限制和质量判断。
skill "promote-skill" {
输入层 -> 用户目标 + 目标文件 + 禁止范围 + 验收标准
上下文层 -> Overview / When to Use / Inputs to collect or infer
规则层 -> SKILL.md 触发条件 / 执行顺序 / 输出格式
运行层 -> 无特殊运行时 | 读取文件、写入/修改文件、执行终端命令、读取环境变量 | 主要在本地完成
安全层 -> 通常不需要额外 API Key + 小任务验证 + diff / 日志复核
输出层 -> 可复制结果 + 检查清单 + 下一步迭代
} Promote Skill
Overview
Use this skill to turn a local SKILL.md into a public marketplace listing and promotion checklist. The workflow is agent-friendly:
- Validate and sanitize the skill source.
- Create a public source URL, preferably a GitHub repo or public gist raw URL.
- Publish to uGig first because uGig imports the raw
SKILL.md, scans it, and exposes a "Publish Everywhere" checklist. - Use sh1pt as the cross-marketplace promotion manifest/command generator.
- Work through other marketplaces by CLI, PR, account submission, or manual upload depending on the platform.
Never include credentials, cookies, private env values, or personal secrets in a public skill file or marketplace listing.
When to Use
Use this when the user says:
- "publish this skill"
- "promote this skill"
- "put it on uGig / ClawHub / LobeHub / Goose / Kilo"
- "make a public SKILL.md"
- "publish everywhere"
- "create a promote-skill workflow"
Do not use this for job applications themselves; use job-board specific skills for that.
Inputs to collect or infer
Required:
- Local
SKILL.mdpath or directory containing it. - Public-safe title, slug, tagline, description, category, tags.
- Price in sats;
0means free.
Optional:
- Existing public raw
SKILL.mdURL. - GitHub repo/gist preference.
- uGig credentials or already-authenticated browser profile.
- Marketplace-specific credentials/API keys for platforms that need them.
Safety checklist before publishing
Run a secrets scan before any public upload:
grep -RInE 'password|passwd|secret|token|api[_-]?key|BEGIN .*PRIVATE|private[_-]?key|cookie|credential' SKILL.md . 2>/dev/null || true
Then manually inspect hits. Mentions of environment variable names are fine; real values are not.
For paid listings, remember: marketplaces that import from a public raw URL may expose the artifact publicly. If the artifact is not intended to be free, verify access-gating from an anonymous browser before claiming it is paywalled.
uGig workflow
Preferred CLI path once ugig supports skills new:
ugig skills new \
--title "My Skill" \
--description "Public credential-free SKILL.md for ..." \
--tagline "Short marketplace tagline" \
--category Automation \
--price 0 \
--tags "skills,automation,agents" \
--source-url "https://raw-or-gist-url/SKILL.md"
ugig skills create is an alias for ugig skills new. If the installed CLI is older or does not support the needed fields, use the uGig browser form:
https://ugig.net/dashboard/skills/new
Known form fields:
- Skill File URL: raw/direct public
SKILL.mdURL. - Title.
- Tagline.
- Description.
- Price:
0for free. - Category.
- Tags: keep combined tags short; supported-agent chips may count toward the maximum.
- ClawHub URL: fill after publishing to ClawHub.
Verification:
- Listing URL is
/skills/<slug>. - Page says
Freeor the expected price. - Security scan says
Clean. - Medium warning for environment variable access is expected if the skill documents env vars.
- Source URL is correct and contains no secrets.
sh1pt workflow
The sh1pt CLI should provide top-level skill promotion commands:
sh1pt skills new --skill-file ./SKILL.md --source-url "https://raw-or-gist-url/SKILL.md" --price 0
sh1pt skills publish --all --dry-run
sh1pt skills publish --marketplace ugig clawhub goose
sh1pt skills marketplaces
Expected behavior:
sh1pt skills newcreatessh1pt.skill.jsonby reading frontmatter from the localSKILL.mdand filling title/slug/description/tags/price/source URL.sh1pt skills publish --all --dry-runprints exact commands or manual steps for every known marketplace.- Non-dry runs should not blindly execute unknown third-party CLIs until credentials/API setup is verified; print commands and mark manual/action-required states.
Implementation locations vary by project. In a typical sh1pt-style monorepo, add the command module under the CLI package and register it from the CLI entrypoint.
Marketplace matrix
| Marketplace | Method | Publish pattern |
|---|---|---|
| uGig | CLI/API or browser | ugig skills new ... --source-url <raw SKILL.md> |
| ClawHub | CLI | clawhub publish . --slug <slug> --version 1.0.0 |
| skills.sh | Auto-indexed | Push public GitHub repo containing SKILL.md |
| LobeHub Skills | Submit | Use site submission; install command shown as npx @lobehub/cli skill install <slug> |
| Goose Skills | PR / install URL | goose skill add <raw SKILL.md URL>; submit PR if directory requires it |
| Kilo Marketplace | PR | Fork + PR with valid SKILL.md; install command kilo skill install <slug> |
| Skillstore | GitHub repo | Submit repo/raw URL for security analysis |
| FreeMyGent | Upload | Upload skill.md, set price, connect wallet |
| ClawMart | API | clawmart publish . --name <slug> |
| Manus Agent Skills | Account | Free account required; submit through account UI |
| VS Code Agent Skills | GitHub | Publish via extension-indexed GitHub repo/PR |
| Moltbook / NormieClaw | Submit | Submit, set price, pass quality check |
Public source choices
Quick gist:
gh gist create ./SKILL.md --public --desc "<skill title>"
gh api gists/<gist-id> --jq '.files["SKILL.md"].raw_url'
Better for auto-indexers:
mkdir -p /tmp/<slug>
cp ./SKILL.md /tmp/<slug>/SKILL.md
cd /tmp/<slug>
git init
git add SKILL.md
git commit -m "Add <slug> skill"
gh repo create <owner>/<slug>-skill --public --source=. --push
Prefer a repo over a gist when targeting skills.sh, VS Code Agent Skills, Goose/Kilo PRs, or marketplaces that require repository metadata.
Common pitfalls
- Using a human gist page instead of the raw URL. Use
gist.githubusercontent.com/.../raw/.../SKILL.mdfor import. - Publishing credentials in examples. Use placeholder env var names only.
- Too many tags on uGig. Keep tags under 10 and leave supported-agent chips blank if validation complains.
- Assuming "Publish Everywhere" means automatic publication. It is a per-marketplace checklist; many require account registration, API keys, or PRs.
- Calling unknown CLIs live without checking auth. Prefer dry-run until
command -v, login status, and target repo/account are verified.
Final report format
Published:
- uGig: https://ugig.net/skills/<slug>
- GitHub source: https://github.com/<owner>/<repo>
- Raw SKILL.md: https://...
Ready/manual next:
- ClawHub: <command or login needed>
- Goose: <PR/command>
- LobeHub: <submission URL>
Security:
- Secret scan: clean / reviewed
- uGig scan: Clean; env-var warning expected
先判断是否适合
作者设计意图
作者的方法与取舍
边界和复核