Agent Skills › AjayIrkal23/agentic-mercy-10x › agent-development

agent-development

GitHub

用于创建、修订或审查自主代理（Agent）的定义、触发文本及系统提示词。明确代理的命名、范围、元数据质量及边界，确保其作为独立自治助手而非普通技能存在。

skills/agent-development/SKILL.md AjayIrkal23/agentic-mercy-10x

Trigger Scenarios

创建新的本地代理定义重写现有代理的系统提示或触发描述评估某功能是否应设计为代理

Install

npx skills add AjayIrkal23/agentic-mercy-10x --skill agent-development -g -y

More Options

Use without installing

npx skills use AjayIrkal23/agentic-mercy-10x@agent-development

指定 Agent (Claude Code)

npx skills add AjayIrkal23/agentic-mercy-10x --skill agent-development -a claude-code -g -y

安装 repo 全部 skill

npx skills add AjayIrkal23/agentic-mercy-10x --all -g -y

预览 repo 内 skill

npx skills add AjayIrkal23/agentic-mercy-10x --list

SKILL.md

Frontmatter

{
    "name": "agent-development",
    "schema": 1,
    "category": "general",
    "surfaces": [
        "general"
    ],
    "triggers": {
        "paths": [],
        "intents": [
            "general"
        ],
        "keywords": [
            "agent",
            "autonomous",
            "creating",
            "definitions",
            "development",
            "helpers",
            "prompts",
            "reusable",
            "revising",
            "system",
            "text",
            "trigger"
        ]
    },
    "platforms": [
        "linux",
        "darwin",
        "windows"
    ],
    "token-cost": 372,
    "description": "ALWAYS invoke when creating or revising reusable agent definitions, trigger text, or system prompts for autonomous helpers.",
    "disable-model-invocation": true
}

Agent Development

Use When

Creating a new local agent definition.
Rewriting an existing agent prompt, trigger description, or agent metadata.
Reviewing whether an autonomous helper should be an agent instead of a skill or command.

Do Not Use

Building normal skills or command workflows.
Creating slash-command style helpers.
Implementing product code or business logic.

Owns

Agent naming, scope, and trigger phrasing.
Agent system-prompt shape and guardrails.
Agent metadata quality for reliable routing.

Does Not Own

Command frontmatter and slash-command UX.
General project architecture or feature implementation.
Library-specific docs lookup policy.

Combine With

workflow-orchestrator for multi-agent delivery planning.
External skill-creator when the task is really a skill design task.
tool-and-doc-selection if the agent must rely on docs or repo tools.

Workflow

Confirm the agent's job is autonomous and bounded.
Write a trigger-focused description that matches realistic user prompts.
Keep the system prompt narrow, opinionated, and outcome-based.
Point to supporting references or scripts when they exist instead of embedding long manuals.
Validate the agent against one or two realistic prompts.