Discover gists

OpenClaw Prompts - Build Your Own AI Assistant

Prompts to recreate each piece of the OpenClaw system. Use these with any AI coding assistant.

1. Personal CRM "Build a personal CRM that automatically scans my Gmail and Google Calendar to discover contacts from the past year. Store them in a SQLite database with vector embeddings so I can query in natural language ('who do I know at NVIDIA?' or 'who haven't I talked to in a while?'). Auto-filter noise senders like marketing emails and newsletters. Build profiles for each contact with their company, role, how I know them, and our interaction history. Add relationship health scores that flag stale relationships, follow-up reminders I can create, snooze, or mark done, and duplicate contact detection with merge suggestions. Link relevant documents from Box to contacts so when I look up a person, I also see related docs."

2. Meeting Action Items (Fathom)

Getting wget for Windows

For most of the Unix-like work I do on Windows, the small set of utilities provided by Git for Windows suffice. But just yesterday I found myself needing wget to complete a small (Windows) shell script I wanted to write.

In my search for a solution I finally took a closer look at the Git for Windows SDK, and lo! found it includes a wget binary for Windows.

There are a few ways I could have gone with this. The "right" way would have been to replace my existing install with a recompiled Git for Windows. But I wasn't going to exert that kind of effort to get a single utility. What I did instead was install wget for the SDK and then copy it and its one dependency to where I had Git installed.

First install the SDK and open its shell (msys1_shell.cmd). Then install wget using pacman(pacman -Sy wget).

Next, open a Windows terminal as Admin and copy the wget executable and

OpenClaw after 50 days: all prompts

Companion prompts for the video: OpenClaw after 50 days: 20 real workflows (honest review)

These are the actual prompts I use for each use case shown in the video. Copy-paste them into your agent and adjust for your setup. Most will work as-is or the agent will ask you clarifying questions.

Each prompt describes the intent clearly enough that the agent can figure out the implementation details. You don't need to hand-hold it through every step.

My setup: OpenClaw running on a VPS, Discord as primary interface (separate channels per workflow), Obsidian for notes (markdown-first), Coolify for self-hosted services.

After prolonged use and numerous issues with the Keychron K8 in Bluetooth mode, I concluded that the problem was entirely hardware-related after exhausting all possible software solutions.

Upon opening the keyboard, I noticed that the Bluetooth module was placed in a highly unsuitable location—sandwiched between the aluminum case on top and the battery underneath. This positioning severely impacted signal strength, leading to frequent connection drops and input issues.

To improve connectivity, I decided to relocate the antenna by extending it using a blue wire from a Cat6 Ethernet cable, which was the best option available at the time. While the wire’s dimensions and properties might not be ideal, a Wi-Fi antenna from an old laptop or router would likely be a better choice.

Ref: https://rnsaffn.com/poison3/

Functions are set up for Netlify, but should work with some refactoring on other platforms.

pf function, this function serves up poisoned data to anyone that visits (/.netlify/functions/pf)

import type { Handler } from '@netlify/functions';

const URL = 'https://RNSAFFN.com/poison2/';

To enable Poison Fountain for Apache on RHEL/CentOS, copy the poison_fountain.conf file in this gist to /etc/httpd/conf.d/poison_fountain.conf (if you've already configured mod_proxy then omit the two LoadModule lines).

Then tell SELinux that apache is allowed to make outbound http requests by running:

setsebool -P httpd_can_network_connect on

Then restart apache:

	"""
	The most atomic way to train and run inference for a GPT in pure, dependency-free Python.
	This file is the complete algorithm.
	Everything else is just efficiency.

	@karpathy
	"""

	import os # os.path.exists
	import math # math.log, math.exp

	#!/bin/bash

	# ==============================================================================
	# Debian/Ubuntu 代理服务器优化脚本 (v2025.12-fixed)
	#
	# 适用场景: <100 用户的代理服务及小型网站，注重弱网环境下的性能与稳定性。
	# 核心优化: 开启 BBRv1, 优化 TCP 参数, 调整文件句柄, 禁用 IPv6。
	# 特点: 安全、幂等 (可重复运行)、智能验证、有详细注释。
	# 修复: 自动加载 nf_conntrack 模块，避免 sysctl 参数应用失败
	# ==============================================================================

	import numpy as np
	import cv2

	img = cv2.imread('baseball.png')
	imgray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
	ret, thresh = cv2.threshold(imgray, 127, 255, 0)
	contours, hierarchy = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE)
	print("Number of contours = " + str(len(contours)))
	print(contours[0])