I've just started using DeepSeek Harness and I'm curious about other people's experience with it. Two things in particular: Are you running into any issues — reliability, tool use, following instructions in agents.md, anything else? How does it compare to Claude Code and Codex in real use? Especially on multi-file projects where the agent has to keep the whole codebase consistent.Thanks in advance

🔥 Has trying new AI tools made you more productive or just busier? My current take: Everyone has their own versions of things, if you try to learn them all definitely going to be busier. But if you are aware of more things than you actually have a deep knowledge on that's the way. Example I'm aware of muse code but haven't tried it yet. New tools can unlock better workflows, but constantly switching also resets the learning curve. Sometimes the breakthrough comes from going deeper with the tool already open, not adding another subscription. Which describes you right now? 💬 A: Exploring is helping me discover better workflows B: I need to stop switching and master my current stack C: I’ve found a healthy balance Reply with A, B, or C and name the tool that has earned a permanent place in your workflow. If someone names a tool you use, compare what each of you does with it 📝.