Prompts are technical debt too

Adding code is necessary to deliver new features, but each line increases complexity and maintenance costs. Future changes must work with existing code or avoid breaking it, and accumulated code can become impossible for one person to fully understand, forcing reliance on guesses and heuristics. Sensible engineering aims to write as little code as possible. For AI-based programs, prompts become a key lever, often organized into codebase-specific prompt files and tool- or capability-specific instructions. Small prompt tweaks can produce large performance gains, and differences across AI tools typically come from subtle prompting variations. Prompt changes also occur when switching tools, workflows, adding loops, skills, or external servers, even without writing new code.

"It’s common and correct to say that “all code is technical debt”. Adding code is a necessary evil for developing new features: you almost always have to do it, but each line of code adds to the complexity and maintenance burden of the system. All future changes to the system have to work with the existing code, or at least avoid breaking it. Once systems accumulate enough code, they become impossible for a single person to understand: instead of reading the code and understanding what it does, you must rely on guesses, theories and heuristics. Sensible engineers write as little code as possible."

"They write a lot of prompts, though! Many large projects now have a set of codebase-specific prompt files: AGENTS.md, CLAUDE.md, those same files in sub-directories, and skills. If you’re building a program that uses AI, you’ll have separate prompts for capabilities and for each tool, as well as a whole set of system prompts."

"Prompts are important. Minor tweaks to a LLM’s prompt can unlock significant performance improvements. If the same model feels different across Codex, Cursor, OpenCode, and Copilot, it’s almost certainly due to subtle differences in prompting. AI companies spend a lot of time testing and tweaking their prompts, so it makes sense why engineers would spend a lot of time tweaking their AGENTS.md files for their projects. I’d even call switching tools or workflows to be a form of prompting."

"If I start wrapping my agents in a Ralph loop, pull in a new skill file, or install an MCP server, that’s still a change to my prompts even though I’m not the one who wrote it. I think it is a bad idea to spend a ton of time tweaking a bespoke agentic coding setup. Why is that, given that prompt adjustments can deliver a lot of value? Because prompt adjustments are model-specific. Earlier I said that AI companies spend a lot of time tweaking their prompts. In fact, they spend that amount of time for each new model release."

Read at Seangoedecke