Resources
Playbooks, guides, and best practices for AI-native E2E testing.
End-to-End Testing: What It Is and What It Costs
E2E tests exercise the running application the way a user does, which is why they catch what unit tests cannot and why they have always been expensive to keep. The expensive part is maintenance, and that is the part that recently changed.
GitHub MCP Server: What It Does and Where It Stops
Connecting a coding agent to GitHub removes a lot of copy-paste: issues, pull requests, CI status, reviews. It does not tell the agent whether the code in that pull request works.
MCP Inspector: Debugging an MCP Server Before an Agent Uses It
When an agent cannot use your MCP server, the agent is the worst place to debug it. Inspector connects to the server directly so you can see the tool list, call tools by hand, and read the actual error.
MCP Security: What Changes When a Model Decides
The protocol is not the risk. The risk is that a tool with real permissions is now invoked by something that chooses on its own when to invoke it, based on text it read.
MCP vs API: What Is Actually Different
Most MCP servers wrap APIs, so the two are not alternatives. The real difference is who does the discovering: an API expects a developer to read docs and write integration code, while MCP lets the model find and call the tool at runtime.
OpenCode vs Claude Code: Open Source Agent or Managed One?
OpenCode is infrastructure you point at any model. Claude Code is a managed agent tied to one. The decision is about who owns the harness, and it has consequences for cost, portability, and what you can run offline.