| name | research |
| description | Research a topic on the web and save a structured note to the resources folder. Use when the user wants to investigate a topic from scratch with no specific URL — 'research X for me', 'what do I need to know about Y'. Do NOT use for: processing a specific URL the user already has (use summarize), processing a book already read (use reading-notes), or quick note capture without web research (use capture or braindump). |
| schedulable_with_args | true |
| required_args | ["topic"] |
Research
Research a topic and save the findings as a structured note in your resources folder.
Usage: /research [topic]
Step 1: Clarify the Research Goal
If topic is provided, confirm scope. If not, ask:
What do you want to research?
Then ask:
What are you trying to figure out? (This helps me focus the research.)
Optional: ask depth preference:
- Overview : broad understanding, key points
- Deep dive : comprehensive, with sources and nuance
- Practical : focused on how-to and actionable takeaways
Step 2: Conduct Research
Search for information on the topic. Look for:
- Authoritative sources (docs, papers, established publications)
- Multiple perspectives if the topic is contested
- Practical examples or case studies
- Recent developments (note dates of sources)
Step 3: Synthesize
Before writing the note, synthesize what you found:
- What's the core answer to the user's question?
- What are the key concepts to understand?
- What's actionable or immediately useful?
- What's uncertain or contested?
Step 4: Choose Subfolder
- Glob existing subfolders in
[resources_folder]/*/
- Suggest a kebab-case subfolder based on the research topic (max 2 levels, e.g.
technology/ai)
- Present to user: "I'd file this under
[resources_folder]/[suggested-path]/. OK?"
Show existing subfolders as options.
- Use confirmed path for file creation.
Step 5: Create Research Note
File: [resources_folder]/[subfolder]/[Topic Name].md (subfolder confirmed in Step 4)
If a note on this topic already exists (search recursively in [resources_folder]/**/*.md), ask whether to create a new one or append a "Research : [Date]" section.
---
tags: [research, topic-tag]
created: YYYY-MM-DD
source: /research
sources: [list of key sources]
---
# [Topic Name]
> **Research goal:** [What the user was trying to figure out]
## Overview
[2-3 sentence summary]
## Key Concepts
### [Concept 1]
[Explanation]
### [Concept 2]
[Explanation]
## Key Takeaways
- [Actionable insight 1]
- [Actionable insight 2]
- [Actionable insight 3]
## Open Questions
- [Something the research didn't fully resolve]
## Sources
- [Source 1 : title and context]
- [Source 2 : title and context]
## Related
[[Link to related vault notes]]
Step 6: Suggest Follow-Up
After creating the note:
──────────────────────────────────────────────────────────────
🔍 Research — {topic}
──────────────────────────────────────────────────────────────
Saved to [resources_folder]/[subfolder]/[Topic Name].md
You might also explore:
• {related topic 1}
• {related topic 2}
(omit "You might also explore" block if no follow-up suggestions)
→ Run /summarize [url] to go deeper on a specific source.
→ Run /learn to save key insights to memory.
Progress reporting
This skill is long-running. Emit a 1-line status update after each major step so the user can see progress in real time.
In-session format:
→ [step N/M] <action being taken>
Examples:
→ [step 1/6] fetching seed sources...
→ [step 2/6] expanding queries via search...
→ [step 3/6] reading 12 web sources...
→ [step 4/6] extracting key claims + cross-referencing...
→ [step 5/6] drafting resource note...
→ [step 6/6] saving to 04-resources/ + adding links...
Rules:
- Emit one line per major step (NOT per sub-step or tool call)
- M = total steps known up front (count them before starting)
- Status lines use
→ [step N/M] prefix exactly so they're visually distinct from skill output
- Do NOT emit heartbeats for fast operations (< 5 seconds)
Known Gotchas
-
Versioned tools and libraries. Web search may return cached or outdated documentation pages. Include the version number in search queries when the topic is a versioned library or framework, and note the publication date of each source in the note.
-
No web search tool in context. If no search tool is available, be explicit about which parts of the research come from training data vs. live sources. Do not silently present training knowledge as web-researched findings.