northseadl/norix-skills / llm-agent-dev
llm-agent-dev
UnverifiedLLM Agent engineering: pattern selection (12-mode matrix), data simulation, convergence iteration. Covers intent routing, function calling, ReAct, MCP, prompt chaining, guardrails, evaluation.
evals/evals.json
View on GitHub{
"skill_name": "llm-agent-dev",
"version": "3.0.0",
"evals": [
{
"id": 1,
"prompt": "设计一个电商智能客服 Agent,目标是提升售后自助率。",
"should_trigger": true,
"expected_output": "触发技能;输出包含 Stack/Architecture/Safety/Evaluation/Data&Iteration 五段。推荐 CREX + Slot + Workflow"
},
{
"id": 2,
"prompt": "帮我写 pre-push hook,限制 develop 合并到 main。",
"should_trigger": false,
"expected_output": "不触发;这属于仓库治理技能范围"
},
{
"id": 3,
"prompt": "我们要做点餐 Agent,语音入口,端到端延迟要求 1 秒内。",
"should_trigger": true,
"expected_output": "优先推荐低延迟栈(CREX + Slot),禁用 ReAct/Orchestrator"
},
{
"id": 4,
"prompt": "我要做一个需要跨部门审批的退款流程 Agent。",
"should_trigger": true,
"expected_output": "推荐 Workflow + Handoff,明确 HITL 触发条件和审批节点"
},
{
"id": 5,
"prompt": "我只想要一个普通 FAQ 页面信息架构建议。",
"should_trigger": true,
"expected_output": "可触发,给出轻量化 CREX 方案,不应过度设计多模式栈"
},
{
"id": 6,
"prompt": "请直接给我一个分支策略模板。",
"should_trigger": false,
"expected_output": "不触发;不是 Agent 架构设计任务"
},
{
"id": 7,
"prompt": "请给我一个客服 Agent 的评估闭环设计。",
"should_trigger": true,
"expected_output": "应包含三层评估金字塔、质量门禁、回归测试、持续改进飞轮"
},
{
"id": 8,
"prompt": "把这个项目代码全部重构成多 Agent 架构。",
"should_trigger": true,
"expected_output": "应先做 tradeoff 评估,不得默认 multi-agent 是最优"
},
{
"id": 9,
"prompt": "帮我给这个 Agent 生成一批评估数据集,要覆盖边界情况。",
"should_trigger": true,
"expected_output": "触发 Pillar II 数据模拟,按维度生成 happy_path/edge_case/adversarial 数据集"
},
{
"id": 10,
"prompt": "我有三个版本的 System Prompt,不知道哪个效果最好。",
"should_trigger": true,
"expected_output": "触发 Pillar III 收敛迭代流程,设计多版本 A/B 测试并行评估方案"
},
{
"id": 11,
"prompt": "我需要一个能生成营销文案并自动迭代质量的 Agent。",
"should_trigger": true,
"expected_output": "推荐 Evaluator-Optimizer 模式 (⑫),明确 Generator/Evaluator 分工"
},
{
"id": 12,
"prompt": "设计一个知识库问答 Agent,支持从多个数据源检索。",
"should_trigger": true,
"expected_output": "推荐 Parallelizing + RAG,加载 knowledge-qa 场景蓝图"
},
{
"id": 13,
"prompt": "如何设计 Agent 的上下文工程策略?",
"should_trigger": true,
"expected_output": "加载 context-engineering.md 参考,讲解上下文预算管理和策略"
},
{
"id": 14,
"prompt": "帮我批量生成一套 Agent 评估测试数据,要覆盖边界和对抗场景。",
"should_trigger": true,
"expected_output": "触发 Pillar II 数据模拟生成,Agent 自主调 LLM API 分维度批量生成数据集"
},
{
"id": 15,
"prompt": "让几个 Agent 辩论一下这个架构方案。",
"should_trigger": false,
"expected_output": "不触发;多 Agent 辩论不在此技能范围内"
},
{
"id": 16,
"prompt": "把 PRD 拆成多个任务给 Codex 执行。",
"should_trigger": false,
"expected_output": "不触发;跨 Agent 任务拆解执行不在此技能范围内"
},
{
"id": 17,
"prompt": "我的 Agent 需要防御 Prompt Injection 攻击。",
"should_trigger": true,
"expected_output": "加载 safety-guardrails.md,讲解 5 层安全架构和注入防御策略"
},
{
"id": 18,
"prompt": "设计一个编程助手 Agent,类似 Cursor 那种。",
"should_trigger": true,
"expected_output": "加载 coding-assistant 场景蓝图,推荐 ReAct + Eval-Opt 组合"
},
{
"id": 19,
"prompt": "我需要迭代优化 Agent 的 Prompt,怎么做比较科学?",
"should_trigger": true,
"expected_output": "触发 Pillar III 收敛迭代 + 加载 prompt-patterns.md 参考"
},
{
"id": 20,
"prompt": "如何监控 Agent 的线上质量和成本?",
"should_trigger": true,
"expected_output": "加载 observability.md,讲解 Tracing/Metrics/Logging 三角和 SLO"
}
]
}Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |