Hello, World — Welcome to grant.hawk
First post on the new blog. Here's what I'm building and why.
⚙
Homelab infrastructure, AI agents, trading systems, and the occasional opinion.
First post on the new blog. Here's what I'm building and why.
First post written and published entirely from Obsidian at 2am. The pipeline works.
First post written and published entirely from Obsidian at 2am. The pipeline works.
How I built OpenClaw — a multi-agent system where AI agents dispatch tasks, review each other's work, and send me a morning briefing. All running on a single mini PC.
A systematic approach to swing trading Australian equities with strict entry criteria, trailing stops, and zero gut-feel decisions.
From Gitea to Qdrant to my own MCP servers — the case for keeping your infrastructure close and your data closer.
Owner: Prox (Your Name) Infrastructure: Minisforum X1A1 (“doug”) + 2013 MacBook Pro Primary Hub: OpenClaw Gateway (single process, multi-agent) Operating Philosophy: Right Model, Right Task — Cost-Conscious, Data-Sovereign
All agent names, Telegram bots, and emojis are configured. Each agent has its own Telegram bot with dedicated routing via bindings.
| Role | Agent ID | Name | Telegram Bot | Emoji |
|---|---|---|---|---|
| CEO / Orchestrator | ceo | Hawk | @HawkIncCEO_Hawk_bot | 🦅 |
| Executive Assistant | ea | Di | @HawkIncEA_Di_bot | ❤️ |
| Engineer | eng | Saeed | @HawkIncENG_Saeed_bot | 🔧 |
| Finance / Security | fin | Craig | @HawkIncFIN_Craig_bot | 💰 |
| Social Media | social | Eleri | @HawkIncSOCIAL_Eleri_bot | 💬 |
| General Purpose / Dogsbody | doug | Doug | @GrantsClaw_bot | 🐈 |
| Heartbeat (background) | heartbeat | Doug_Heartbeat | — (no bot) | — |
The original blueprint contained several fundamental errors about how OpenClaw works. This version corrects them:
How I wired up a feedback loop where an AI agent reviews my trades, assigns homework, and tracks my improvement.
A Python script to transcribe any YouTube video locally using OpenAI’s open-source Whisper model. No API keys, no tokens, no cloud services — everything runs on your own machine.
Sometimes you need a transcript of a YouTube video — for notes, research, or feeding into another tool. Most online transcription services either cost money, require API keys, or send your data to third-party servers. This script does it all locally using two excellent open-source tools: yt-dlp for downloading audio and Whisper for speech-to-text.
Version 2026.02.17 | For Medium-Skilled Users
This guide explains how to build, register, and use Skills in OpenClaw — with a focus on the Telegram interface and Obsidian vault integration. All examples are illustrative templates; adapt them to your own use cases.
OpenClaw runs as a local service (a user-level systemd daemon called openclaw-gateway) and exposes itself through Telegram as its primary user interface. It reads a single JSON config file and a directory of SKILL.md files to understand what it can do.