Skill Auto-Use
Stop waiting to be told which skill to use. Match context to skills automatically.
How It Works
Maintain a trigger table in your workspace (e.g., memory/protocols.md or a dedicated skill-triggers.md). The table maps observable contexts to installed skills:
CODEBLOCK0
Rules
1. Every Skill Gets a Trigger
When a skill is installed, add at least one trigger row to the table before doing anything else. No skill should exist without a trigger. If you can't identify a trigger, the skill probably shouldn't be installed.
2. Match Before Asking
On every user message, scan the trigger table mentally. If a skill matches, use it. Don't ask "should I use X?" Just use it. The user installed the skill because they want it used.
3. Multiple Matches Are Fine
If a message matches multiple skills, use all of them. A request about a PDF from a website might trigger both scrapling-official (to fetch it) and markdown-converter (to process it).
4. Audit on Heartbeat
During heartbeat or review passes, check:
- - Are there installed skills without triggers? Add them.
- Are there triggers that never fire? Consider removing the skill.
- Are there repeated manual skill invocations? Add a trigger.
5. Keep Triggers Observable
Triggers should be based on things you can actually detect in the conversation:
- - File types sent (PDF, audio, image)
- Question patterns ("what's the weather", "research X for me")
- Emoji shortcuts (♻️, 🔻, 💮)
- Keywords or domains mentioned
- Task types (deploy, scrape, cook, schedule)
Avoid triggers based on internal state or guesses.
Setting Up
Option A: Add to existing protocols file
Add a ## Skill Auto-Use section to your protocols or memory file with the trigger table.
Option B: Dedicated file
Create skill-triggers.md in your workspace:
CODEBLOCK1
Reference it from MEMORY.md or your session startup sequence.
Template Triggers
Common triggers for popular skills:
| Trigger | Skill |
|---|
| PDF/Word/Excel/HTML file shared | markdown-converter |
| Research question needing multiple sources |
deep-research-pro |
| Web scraping or site data extraction | scrapling-official |
| Meal planning, recipes, grocery lists | feast |
| User correction or mistake identified | self-improving |
| Weather question | weather |
| GitHub issue/PR/CI work | github |
| Audio file shared or transcription needed | openai-whisper-api |
| Image generation requested | nano-banana-pro or openai-image-gen |
| Diagram or flowchart requested | diagram-generator |
| Calendar/email/drive operations | gog |
| Public-facing copy to review | humanizer |
| New skill needed | clawhub |
| Skill installed |
skill-auto-use (add trigger immediately) |
Installation Rule
Add this to your workspace's permanent rules:
When installing any new skill, IMMEDIATELY add it to the Skill Auto-Use trigger table. No exceptions. This is not optional.
技能自动使用
无需等待被告知该使用哪个技能。自动将上下文与技能匹配。
工作原理
在工作区维护一个触发器表(例如 memory/protocols.md 或专门的 skill-triggers.md)。该表格将可观察的上下文映射到已安装的技能:
markdown
| 触发器 | 技能 | 操作 |
|---|
| 用户发送PDF或文档文件 | markdown-converter | 转换为Markdown,总结 |
| 用户提出研究问题 |
deep-research-pro | 多源研究并附引用 |
| 用户纠正你或说那不对 | self-improving | 记录纠正,评估是否推广 |
| 需要网页抓取 | scrapling-official | 使用Scrapling进行获取+解析 |
规则
1. 每个技能都有触发器
安装技能时,在执行任何其他操作之前,至少向表格添加一行触发器。没有触发器的技能不应存在。如果你无法确定触发器,那么这个技能可能不应该被安装。
2. 先匹配再询问
对于每条用户消息,在脑海中扫描触发器表。如果某个技能匹配,就使用它。不要问我应该使用X吗?直接使用即可。用户安装该技能是因为他们希望它被使用。
3. 多个匹配没问题
如果一条消息匹配多个技能,就全部使用。一个关于网站PDF的请求可能会同时触发 scrapling-official(用于获取)和 markdown-converter(用于处理)。
4. 心跳时审计
在心跳或审查过程中,检查:
- - 是否有已安装的技能没有触发器?添加它们。
- 是否有从未触发的触发器?考虑移除该技能。
- 是否有重复的手动技能调用?添加一个触发器。
5. 保持触发器可观察
触发器应基于你在对话中实际能够检测到的事物:
- - 发送的文件类型(PDF、音频、图像)
- 问题模式(天气怎么样、帮我研究X)
- 表情符号快捷方式(♻️、🔻、💮)
- 提到的关键词或域名
- 任务类型(部署、抓取、烹饪、安排)
避免基于内部状态或猜测的触发器。
设置
选项A:添加到现有协议文件
在协议或记忆文件中添加一个## 技能自动使用部分,包含触发器表。
选项B:专用文件
在工作区创建 skill-triggers.md:
markdown
技能触发器
从MEMORY.md或会话启动序列中引用它。
模板触发器
常用技能的常见触发器:
| 触发器 | 技能 |
|---|
| 共享PDF/Word/Excel/HTML文件 | markdown-converter |
| 需要多个来源的研究问题 |
deep-research-pro |
| 网页抓取或网站数据提取 | scrapling-official |
| 膳食计划、食谱、购物清单 | feast |
| 用户纠正或发现错误 | self-improving |
| 天气问题 | weather |
| GitHub问题/PR/CI工作 | github |
| 共享音频文件或需要转录 | openai-whisper-api |
| 请求生成图像 | nano-banana-pro 或 openai-image-gen |
| 请求图表或流程图 | diagram-generator |
| 日历/邮件/云端硬盘操作 | gog |
| 审查面向公众的文案 | humanizer |
| 需要新技能 | clawhub |
| 技能已安装 |
skill-auto-use(立即添加触发器) |
安装规则
将此添加到工作区的永久规则中:
安装任何新技能时,立即将其添加到技能自动使用触发器表中。没有例外。这不是可选的。