skill-creator
Design valid CashCode Skills with concise instructions and progressively loaded scripts, references, templates, and assets.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Design valid CashCode Skills with concise instructions and progressively loaded scripts, references, templates, and assets.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Generate a complete local HTML chart. Use for trends, comparisons, distributions, relationships, flows, or dashboards.
Work with GitHub pull requests, issues, Actions runs, and API data through the gh CLI.
Get current weather and forecasts. Use for temperature, rain, wind, humidity, or forecast questions for a place.
| name | skill-creator |
| description | Design valid CashCode Skills with concise instructions and progressively loaded scripts, references, templates, and assets. |
| version | 1 |
| tags | ["skill","workflow","authoring"] |
| triggers | ["create a skill","update a skill","reusable workflow"] |
| requires | {"tools":["agent_skill_manage"]} |
Create broad reusable workflow packages, not one-turn notes.
^[a-z0-9][a-z0-9._-]{0,63}$.display_name; never put Chinese or other unsupported characters in name.SKILL.md and optional text support files.agent_skill_manage with action=create, the same canonical name, the complete content, and a short reason.success=false, correct the reported validation error and retry the managed tool, or report failure. Never switch to filesystem, shell, or HTTP creation.success=true.Required package shape:
<slug>/
SKILL.md
references/ optional detailed knowledge
templates/ optional editable starters
scripts/ optional deterministic helpers
assets/ optional output resources
Rules:
name; keep localized titles in the optional display_name field.description containing both capability and trigger intent.SKILL.md concise: trigger conditions, ordered steps, pitfalls, and verification.agent_skill_manage with action=create, the canonical slug, the complete SKILL.md, optional text support files, and a short reason.write_file, edit_file, exec, curl, shell commands, or ad hoc HTTP calls.agent_skill_manage result containing success=true as a successful creation. On failure, report its code and detail without claiming the Skill exists.name and a non-empty string description.display_name, version, tags, triggers, always, requires, and optional.display_name is a trimmed string of at most 80 characters without control characters.version is a positive integer and always is a boolean.tags and triggers are lists of non-empty strings.requires and optional are mappings whose tools, mcp_servers, bins, and env values are lists of non-empty strings.references/, templates/, scripts/, or assets/; paths must be package-relative and cannot contain ...Use this shape for a localized title:
---
name: renzhi-niuqu
display_name: 认知扭曲
description: 识别常见认知扭曲并在用户要求检查想法时引导生成平衡表述。
version: 1
tags: [反思, 认知]
triggers: [检查想法, 识别认知扭曲]
always: false
requires:
tools: []
mcp_servers: []
bins: []
env: []
optional:
tools: []
mcp_servers: []
bins: []
env: []
---
# 工作流程
1. 确认用户希望检查的原始想法。
2. 识别适用的认知扭曲类型并说明证据。
3. 生成更平衡、可验证的替代表述。
4. 检查输出是否遵循用户约束。
Validate every package before enabling it. Test scripts explicitly when a Skill introduces them.