After ten or fifteen turns, the original goal gets buried under tool results, and the LLM starts drifting. Goal pinning fixes this structurally: a hook at turn_start re-injects the goal before any tool result lands. The agent physically cannot forget what it cannot unsee. It's the cheapest improvement you can make to a long-running agent, a couple of settings in the workspace.
Engineering notes from the Digitorn team. No marketing, no launch announcements, no "10 prompts that will change your life". Just the things we write that we'd want to read.
