Prompt compiler · v1.0
PromptGen assembles structured prompts from tool profiles, domain rules, depth levels and behavior contracts — then scores and refines them before you ever paste them into a model.
ChatGPT, Claude, Gemini, Perplexity, Copilot, Midjourney, DALL·E, Stable Diffusion, Leonardo and a portable custom profile — each with its own structure, verbosity and formatting rules.
From Minimal to Production-grade and Research-grade. Each level rewrites depth, structure complexity, validation strength and output strictness.
Truth-first, no-guesswork, validation-first, cross-check, risk-aware, compliance-safe and more — compiled into an explicit contract the target model must follow.
Scores your draft, detects vagueness and contradictions, names what is missing, and rewrites weak fields. It asks instead of inventing.
Every save snapshots the full structured draft plus compiled output, so you can compare and restore any earlier version.
One shared prompt schema and one pure compiler — the same core that will power the desktop companion client.
Every prompt is assembled deterministically, in the same order, every time.