Built an open-source tool for sandboxing coding agents
https://t.co/ST0ZqugcFv
Easily spin up and use local environments for Claude Code/Codex/OpenCode/Gemini/etc
Fully free, MIT-licensed
https://t.co/nu2wu7HmYw
Agentic dev tip - when working on your codebase, get the agent to code scripts it can use to test its work.
Think unit tests, but for agents.
This allows the agent to iterate more effectively while working on bugs.
This is a similar idea to code gen vs MCP
AI/Agentic Development Tip - if you're stuck trying to fix a bug, tell the agent to write itself a script and iterate on the solution up to 5 times.
eg - you are calling an API and need to parse the response, but keep getting errors. Get the agent to iterate on it
Want to create game assets with AI?
In this tutorial I share some tips for how to do this with general-purpose AI platforms like Replicate and OpenRouter, rather than game-dev specific services
The focus is on building your own asset pipeline
https://t.co/usrXmrP5cX
@PythagoraAI First I'd learn terminal/shell, then basic apis
after that, three are lots of ways to go
1. databases
2. servers / serverless
3. Deploying
4. building custom APIs
5. MCP servers
And for more advanced:
- load balancing + scaling
- container deployments
- custom model training