Request Research

Create a research request document for: {{input}}

Footnote Format Reference

Read the /research skill for the required footnote format. When generating the request document, bake the footnote format specification inline so external researchers can follow it.

Output

Save to research/requests/[topic-slug].md in the project root.

The document should be usable by any research agent (Claude Code, Claude.ai, or other tools). Assume the agent knows HOW to research — focus on WHAT and WHY.

Important: The researcher has NO access to this repo. Bake in all relevant context — do not link to internal docs. Read relevant project docs first, then include the necessary context directly in the request.

Evidence Standards (CRITICAL)

Research findings are only as valuable as their verifiability. Every significant conclusion needs an evidence chain that can be independently verified.

Required for All Claims

• Direct quotes for key findings (not paraphrases)
• Source links for every significant claim
• Access date for web sources
• Section/page reference for docs (when applicable)

Evidence Quality Hierarchy

Flag Gaps Explicitly

• No authoritative source found → Flag as [UNCERTAIN - no authoritative source]
• Conflicting sources → Present both positions with links
• Outdated sources only → Note recency concern explicitly
• Single-source claims → Note [single source] for transparency

Document Structure

# Research Request: [Topic]

**Created:** [date]
**Status:** Open

## Goal
What we're trying to learn or decide. Be specific about the outcome needed.

## Context
Why this matters. Include all relevant background directly — the researcher cannot access our repo. Explain:
- What project/situation this is for
- What decisions depend on this research
- Any constraints or requirements

## Key Questions
Numbered list of specific questions to answer. These should be:
1. Concrete and answerable
2. Prioritized (most important first)
3. Scoped appropriately (not too broad, not too narrow)

## What Would Change Our Approach
What findings would actually shift our strategy or decisions? This helps the researcher focus on actionable insights rather than general background.

## Known Assumptions to Challenge
Any beliefs we currently hold that should be tested. Be explicit about what we think we know — and why we might be wrong.

## Output Format
What format is most useful? Options:
- Summary with sources
- Comparison table
- Pro/con analysis
- Step-by-step guide
- Annotated source list

## Recency Requirements
[Specify how important recency is for this topic:]
- Critical (2024-2025 only) — fast-moving landscape, older content likely outdated
- Prefer recent — recent sources preferred but older foundational content acceptable
- Not critical — topic is stable, older sources fine

## Instructions for Researcher

### Footnote Format (Non-negotiable)
[Include the footnote format from the skill here - inline reference syntax, footnote definition format with all required components, and naming convention. This must be self-contained for external researchers.]

### Evidence Requirements
- **Quote, don't paraphrase** — Include exact quotes for key findings
- **Link everything** — Every significant claim needs a footnote with source URL
- **Date your sources** — Note publication date AND access date in footnotes
- **Show evidence tiers** — Flag whether sources are Tier 1-4 (see Evidence Standards above)

### Research Process
- Ask clarifying questions before diving in if scope is unclear or assumptions seem off
- Push back if the questions seem misguided or if there's a better framing
- Distinguish between well-established findings and emerging/contested ideas
- Be direct about limitations or gaps in available evidence

### Transparency Requirements
- Flag `[UNCERTAIN]` when no authoritative source exists
- Flag `[CONFLICTING]` when sources disagree (present both sides)
- Flag `[SINGLE SOURCE]` when a claim relies on only one reference
- Flag `[OUTDATED]` when newest source is >2 years old on fast-moving topics

After Creating

1. Create research/requests/ directory if it doesn't exist
2. Inform user where the request was saved
3. Note that findings should go in research/findings/[topic-slug].md when fulfilled

Principles

• Bake in all context — researcher cannot access our files
• Be specific about what decisions this research informs
• Encourage the researcher to push back and ask questions
• Don't over-specify methodology — trust the researcher