clarifylisted

# Tandem Clarify You are a prompt pre-processor. Restructure input into a well-formed prompt — fixing grammar, typos, formatting, and intent clarity — then execute it immediately. This skill is for manual invocation, providing higher-quality restructuring via the main model with the full Prompt Structure Template. The auto-detection hook (`detect-raw-input.sh`) uses haiku for lighter, faster assessment on every prompt above the length threshold. ## Process 1. **Parse** - Extract the actual request, intent, constraints, and desired output 2. **Restructure** - Fix grammar/spelling, clarify intent, add structure. Map onto relevant sections of the Prompt Structure Template (see [references/prompt-structure-template.md](references/prompt-structure-template.md)) 3. **Proceed normally** - The restructured prompt is now a well-formed request. Apply your normal judgment — if it warrants plan mode, enter plan mode. If core intent is genuinely ambiguous, ask one targeted question. Otherwise, execute directly. The restructuring makes the right path obvious; don't second-guess it with extra routing logic ## Transformation Rules Apply these during restructuring: **Wrap reference material in XML tags** — Any pasted content, code, data, or context gets descriptive tags (`<error_log>`, `<api_spec>`, `<existing_code>`). Place long inputs BEFORE the instructions that reference them. **Add the WHY** — When a constraint's motivation is inferrable from context, make it explicit. This drama