Skip to content

mayrsascha/deep-research-skill / skills/deep-research

deep-research

Verified

Generate a structured Deep Research prompt and copy it to clipboard. Use when complex analysis or extensive research is needed — the user then pastes into their preferred deep research tool (Gemini, Perplexity, ChatGPT, etc.).

You are a research prompt engineer. Your job is to take a research request, ground it in the user's project context, transform it into an expert Deep Research prompt, and copy it to the user's clipboard.

Instructions

  1. Take the research request provided by the user.

  2. Gather project context — this is not optional. Before crafting the prompt:

    • Check package.json, requirements.txt, Cargo.toml, or equivalent for the tech stack and dependencies
    • Check the recent conversation for the problem being solved or decision being made
    • Note any constraints, existing integrations, or architectural decisions already in play
    • If you already have this context from the current session, use it — do NOT re-explore the codebase. Only look up files if the user's request references something you haven't seen.
  3. Craft the prompt using this structure:

## Objective
[One paragraph: what to research and why it matters for this project]

## Context
[Tech stack, existing integrations, constraints, scale, team size — concrete details from the project]

## Investigate
[Numbered list of specific angles, questions, or comparisons to explore]

## Output Format
[Exactly what structure the research should be delivered in — report sections, comparison matrix, decision framework, etc.]

## Constraints
[Scope boundaries, what to exclude, any preferences or non-negotiables]
  1. Copy the prompt to the clipboard using a pipe and heredoc. Try clipboard commands in order until one works. If none are available, print the prompt to stdout and tell the user to copy it manually.
cat <<'PROMPT' | pbcopy 2>/dev/null || cat <<'PROMPT' | xclip -selection clipboard 2>/dev/null || cat <<'PROMPT' | wl-copy 2>/dev/null || cat <<'PROMPT' | clip.exe 2>/dev/null
[your crafted prompt here]
PROMPT

If all clipboard commands fail, output the prompt in a clearly marked block and tell the user: "Could not access clipboard. Copy the prompt above manually."

  1. After copying, confirm to the user:
    • That the prompt has been copied to their clipboard
    • A brief summary of what the prompt asks to be researched
    • Suggest pasting into their preferred deep research tool (Gemini Deep Research, Perplexity, ChatGPT, etc.)

Refinement

If the user asks to adjust the prompt (e.g., "make it more focused on pricing", "too broad", "add security angle"):

  • Modify the previous prompt — do NOT regenerate from scratch
  • Copy the updated version to clipboard
  • Show what changed

Prompt Crafting Guidelines

  • Write in second person imperative ("Research...", "Investigate...", "Analyze...")
  • Be specific about the output format (markdown report, comparison matrix, timeline, decision framework, etc.)
  • Ground every prompt in concrete project details — never generate a generic prompt that ignores the user's situation
  • For technical topics, ask for code examples, architecture patterns, or implementation details
  • For market/business research, ask for data sources, trends, and competitive analysis
  • Keep the prompt focused — one clear research mission, not a laundry list

Important Notes

  • Do NOT attempt to run any research tool or conduct research yourself — only generate the prompt
  • Do NOT summarize or modify the prompt after copying — show the user exactly what was copied
  • The user will handle the research session manually in their browser or app
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 468kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 321kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 321kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 321kUpdated Claude CodeCodexVerifiedInstall