メニュー
Create, update, or manage universal-ai-config agent templates. Handles finding existing agents, deciding whether to create or modify, and writing the template.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/fabis94/universal-ai-config --skill update-agentコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
A skill used to exercise preserveOnClean against skill output
Create, update, or manage universal-ai-config instruction templates. Handles finding existing instructions, deciding whether to create or modify, and writing the template.
First example skill for add-skill tests
SKILL.md を表示中
SOC 職業分類に基づく
| name | update-agent |
| description | Create, update, or manage universal-ai-config agent templates. Handles finding existing agents, deciding whether to create or modify, and writing the template. |
| userInvocable | false |
Agents are specialized AI personas with scoped tools and permissions that run in isolated contexts.
Important: Agents are supported by Claude, Copilot, and Codex. Cursor does not support agents — consider using a skill with forkContext: true as an alternative for Cursor. Codex emits each agent as a standalone .codex/agents/<name>.toml file (body becomes developer_instructions); see the "Codex caveats" section in <%= instructionPath('uac-template-guide') %>.
List files in <%= agentTemplatePath() %>/ to discover existing agent templates. Read their frontmatter to understand each agent's purpose and capabilities.
.md file in <%= agentTemplatePath() %>/ with a descriptive name (e.g. code-reviewer.md)name and descriptionSee the Agents section in <%= instructionPath('uac-template-guide') %> for the complete field reference and per-target override syntax. Key fields: name, description, model, tools, permissionMode, hooks, plus Codex-specific nicknameCandidates and sandboxMode.
When targeting Codex alongside Claude/Copilot, three fields require special attention because their vocabularies don't align (see "Cross-target value gotchas" in <%= instructionPath('uac-template-guide') %>):
model: Use a per-target override. Claude takes claude-* model IDs; Codex takes gpt-5.4, gpt-4.1, etc.permissionMode vs sandboxMode: permissionMode is Claude-specific (acceptEdits, bypassPermissions, plan). For Codex, set sandboxMode: { codex: "workspace-write" } (values: read-only, workspace-write, danger-full-access).tools: Codex has no agent-level tool restriction — configure per-server enabledTools on the MCP server config instead. Universal tools is dropped for Codex with a warning.---
name: security-reviewer
description: Reviews code for security vulnerabilities and best practices. Use proactively after code changes.
tools: ["Read", "Grep", "Glob", "Bash"]
model: sonnet
---
You are a security-focused code reviewer. When invoked:
1. Identify the changed files
2. Check for common vulnerabilities (injection, XSS, auth issues)
3. Review dependency usage for known CVEs
4. Report findings by severity (critical, warning, info)