PromptsOpenClaw
A prompting guide for the model you actually run
Use case
OpenClaw is a pile of markdown instructions, and each model follows them differently. You keep one local guide for the model you use. Any time a skill or a prompt file changes, that edit is checked against the guide first.
What to connect
Nothing external. The guide is a markdown file in the workspace.
Setup
- Set MODEL_NAME to the AI model this assistant uses most. If you switch AI models later, say so and regenerate the guide.
- Ask OpenClaw to read the public prompting notes from the lab that ships that AI model, then write [PROMPT GUIDE PATH] in your own words.
- Point every skill-editing flow at that file.
- Change one existing skill on purpose and confirm the edit cites the guide.
Edit before you send: The model name and the path of the guide.
Prompt to paste into OpenClaw
Create a prompting guide for [MODEL NAME] at [PROMPT GUIDE PATH]. Read the public prompting notes from the lab that ships this model, then write the guide yourself. Do not paste their pages in. The guide should say, with a reason for each line: - Shouting with urgency words in all capitals makes this model over-trigger. Write the rule in a normal sentence and say why it exists. - A rule sticks better when the reason is next to it, because the model can generalize from the reason. - Show the shape of a good result. Leave out examples of the failure mode, because the model sometimes copies the failure. - Do not add "if in doubt, call this tool." That line makes tools fire when they should not. - Format the prompt the way you want the answer formatted. Whenever you create or edit a skill, a system prompt, or any other markdown instruction, read this guide first and follow it. If a change would break one of these lines, say which line and wait. If I switch models, rewrite the guide for the new model. Keep the old file as [PROMPT GUIDE PATH].previous so I can diff them.
More OpenClaw prompts
- Personal CRM from mail and calendar
A year of email and meetings is a pile of names you cannot search. OpenClaw reads Gmail and Google Calendar, drops newsletters and cold pitches, and keeps a local profile for each person worth remembering. You ask in plain language who you know at a company, or who has gone quiet.
- Meeting action items with an approval queue
Meetings end and the promises disappear. OpenClaw waits until a meeting is over, pulls the Fathom transcript, matches the room to your CRM, and sends you each action item to approve. Approved items become Todoist tasks. Things other people owe you stay on a waiting-on list, except people on your own team.
- Urgent email, only when it can wait no longer
You do not live in the inbox, including on weekends. OpenClaw checks mail on a timer, ignores senders you have already marked as noise, and pings a Telegram topic only when something needs you now. When you say an alert was wrong, the next check is stricter.
- Knowledge base you fill from a Telegram topic
Links you meant to remember are scattered across chats. You drop a URL into one Telegram topic. OpenClaw saves the article, the video transcript, the full post thread, or the PDF, then you ask later in plain language. A new save can point at something you already kept.