研究与学习
想法漏斗自动推进(Idea Pipeline)
把白天收集的想法,凌晨自动做调研、可行性评估,再写回任务系统。
目录
这个案例能帮你做什么
- 不再让想法“只采集不研究”,每天早上直接看到可执行结论。
- 每条想法自动补齐:竞品/开源替代、技术复杂度、预估工期、主要阻塞。
- 默认限制每晚只处理优先级最高的 3 条,避免信息过载。
你需要的 Skills(按类型)
| 类型 | Skill / 工具 | 用途 | 来源 |
|---|---|---|---|
| 内置 | web_search |
研究市场与技术方案 | OpenClaw Built-in |
| 内置 | 子代理能力(sessions_spawn) |
多想法并行研究 | OpenClaw Built-in |
| 外部 | 任务系统 API(Todoist / Trello / Asana) | 写回研究结果 | 对应平台 |
| 渠道 | Telegram / Discord / Email / 文件目录 | 作为想法采集入口 | 用户自选 |
快速体验版(先跑一轮)
先手工触发一次,验证全链路:
openclaw cron run idea-pipeline
稳定自动版(可长期运行)
1) 放入 cron.jobs(原文配置)
{
"name": "idea-pipeline",
"schedule": {
"kind": "cron",
"expr": "0 3 * * *",
"tz": "UTC"
},
"payload": {
"kind": "agentTurn",
"message": "Process ideas from [YOUR_CAPTURE_METHOD]. For each idea: 1) Search web for existing solutions and market landscape. 2) Analyze technical feasibility (complexity 1-5, estimated time, blockers). 3) Write summary to [YOUR_TASK_SYSTEM] with findings, top 3 similar solutions, feasibility assessment, and next steps. Be concise."
},
"sessionTarget": "isolated"
}
把以下占位符换成你的真实入口:
[YOUR_CAPTURE_METHOD]:例如Telegram bot/Discord #ideas[YOUR_TASK_SYSTEM]:例如Todoist/Trello
2) 工具最小配置(原文示例)
tools:
web_search: {}
message: {}
todoist: {}
3) 详细执行提示词(原文)
Process the ideas captured from [YOUR_CAPTURE_METHOD].
For each idea:
1. SEARCH: Search web for
- Existing products/services solving similar problem
- Open source alternatives
- Technical approaches others used
- Market size or demand indicators
2. ANALYZE feasibility:
- Technical complexity (1-5 scale)
- Estimated build time
- Key technical decisions needed
- Potential blockers
3. WRITE summary to [YOUR_TASK_SYSTEM]:
- Brief overview of findings
- Top 3 similar solutions found with links
- Feasibility assessment
- Recommended next steps
Be concise. Focus on actionable insights, not exhaustive research.
Limit to top 3 ideas per night by priority.
4) 并行增强(可选)
for (const idea of ideas) {
sessions_spawn({
task: `Research: "${idea.text}". Search web, analyze feasibility, return structured summary.`,
agentId: "researcher",
model: "gemini"
});
}
成功标准
- 每天都有稳定的“研究后想法”输出,而不是只堆积原始灵感。
- 每条想法都包含复杂度、工期、阻塞与下一步。
- 晚间处理量受控(Top 3),信息密度高但不拥挤。