prompt engineering automation
i use this for consultancy — any client prompt engineering task runs through this pipeline before delivery.
the problem
prompt engineering is skill-intensive and repetitive. analysing the task, writing the spec, drafting and validating two prompts — done well, this takes hours per client. there was no systematic way to apply senior-level prompt engineering consistently across multiple clients without sacrificing quality.
the approach
- 01 planner
analyses the client task and produces a structured specification — audience, tone, constraints, success criteria, and edge cases. you review and optionally add feedback before anything moves forward.
- 02 system prompt generator
builds a production-grade instruction set for the ai model from the specification. feedback from stage 1 is injected into this stage's input.
- 03 task prompt generator
writes the end-user-facing prompt or prompt chain, tested against both the spec and the system prompt. the three artefacts are now complete.
- 04 validator
formal quality check across all artefacts against prompt engineering principles. returns a verdict — ready to ship, or needs revision with specific notes. approved runs are saved to history.
the workflow
