prompt-guidelineslisted

# Prompt Guidelines Apply prompt-engineering practices when authoring or reviewing any instruction artifact a coding agent will consume: task prompts, feature specs, directives, recipe steps, skill bodies, and review instructions. ## Workflow 1. Identify the prompt artifact and who consumes its output (a human or a script). 2. Triage rigor to blast radius: one sentence for trivial edits, a full task card for multi-file, behavioral, or risky work. 3. Read [the complete guidelines](references/GUIDELINES.md), selecting the sections relevant to the artifact. 4. Anchor to exact files, symbols, tests, and commands from the target project; frame bugs as current vs. expected behavior. 5. Define boundaries (allowed, forbidden, ask-before), validation commands, stop conditions, and the output contract. 6. Do not mutate the project during a review-only request. 7. Check the finished prompt against the pre-send checklist in the guidelines. ## Output Summarize the rules applied, the prompt produced or the findings, exceptions, and validation evidence.