Transform a user's draft prompt into a long-horizon coding-agent task prompt. Use when the user asks to make a prompt long-horizon, turn something into a long-running agent task, add long-horizon scaffolding, or invokes "$long-horizon" or "/long-horizon". Appends framing for running far without unnecessary stops, using subagents for scoped background work, and maintaining implementation-notes.html with decisions, deviations, tradeoffs, and open questions.
Create polished static HTML walkthrough, onboarding, explainer, or concept pages from Markdown, source files, or a short interview. Use when the user wants to turn .md docs into a good-looking HTML page; create a useful human-facing web page for understanding a codebase, workflow, design, paper, CLI, evaluation concept, or project onboarding; or asks for a walkthrough page, onboarding page, interactive explainer, one-page demo, copyable command guide, or concept map. Starts with adaptive interview checkpoints, then builds a single-file static page with accurate commands, expected outputs, diagrams or inspectors, and validation.
Get an independent second opinion from a separate AI model on code, specs, diffs, PRs, or GitHub issues. Use when the user asks for "second opinion", "adversarial review", "independent review", or "codex review". Supports subagent (fresh-context sub-Claude, default), interactive (tmux + codex), codex exec, claude -p, and AWS Bedrock backends. Arguments: [target] [--backend subagent|interactive|codex|claude|bedrock] [--model name] [--effort low|medium|high] [--quick]
Deep project interview that produces actionable specs before implementation begins. Conducts a structured requirements interview with adaptive depth and checkpoints, then outputs a spec as a file, GitHub issue, or both. When given an existing plan or design, switches to grill mode — stress-testing decisions and walking the design tree. Use when the user wants to scope a project, plan a feature, write a spec, stress-test a design, or says "interview" or "grill me".
Extract a DDD-style ubiquitous language glossary from the current conversation, flagging ambiguities and proposing canonical terms. Saves to UBIQUITOUS_LANGUAGE.md. Use when the user wants to define domain terms, build a glossary, harden terminology, create a ubiquitous language, or mentions "domain model" or "DDD".