Research

Conduct automated research on a note, gathering competitor information, relevant discussions, and market signals. Append findings to the note.

Arguments

$ARGUMENTS - The note to research.

Target: Can be:

• A note filename (e.g., voice-notes-app or voice-notes-app.md)
• A relative path (e.g., notes/voice-notes-app.md)
• An absolute path

Examples:

/vault:research voice-notes-app
/vault:research notes/car-search/prd.md
/vault:research ./my-notes/cool-project.md

Subagent Strategy

Research uses subagents to parallelize searches:

Task	Model	Why
Competitor search	haiku	Simple web searches, fast
Discussion skimming	haiku	Reading Reddit/HN threads
Deep competitor analysis	sonnet	Nuanced positioning analysis
Market synthesis	sonnet	Combining multiple sources

Parallel execution: Launch competitor, discussion, and market searches simultaneously as background tasks, then synthesize results.

Process

Step 1: Load the note

Read the specified note. Extract:

• The core topic/problem
• Target audience (if mentioned)
• Any existing research
• Current tags

If the note is too vague, ask for clarification before proceeding.

Step 2: Launch parallel research

Launch 3 subagents in parallel:

Task(
  subagent_type: "search-specialist",
  model: "haiku",
  prompt: "Find competitors for [topic]...",
  run_in_background: true
)

Task(
  subagent_type: "search-specialist",
  model: "haiku",
  prompt: "Find Reddit/HN/forum discussions about [topic]...",
  run_in_background: true
)

Task(
  subagent_type: "search-specialist",
  model: "haiku",
  prompt: "Find market size, funding news, trends for [topic]...",
  run_in_background: true
)

Wait for all to complete using TaskOutput.

Step 3: Competitor research

Search for:

• Direct competitors (products solving the same problem)
• Adjacent products (related solutions)
• Failed attempts (products that tried and shut down)

Search queries:

• "[core topic] app" or "[core topic] software"
• "[core topic] startup"
• "best [solution type] tools 2025"
• Product Hunt search for the category

For each competitor found:

• Name and URL
• Brief description
• Pricing (if visible)
• Key differentiators

Step 4: Discussion research

Search for community discussions:

• site:reddit.com "[topic]"
• site:news.ycombinator.com "[topic]"
• "[topic]" forum OR discussion

Look for:

• Pain points people express
• Solutions they've tried
• What's missing from current options
• Willingness to pay signals

Use WebFetch to read 2-3 most relevant discussions.

Step 5: Market signals

Search for:

• "[topic]" market size
• "[competitor]" funding OR acquired
• Recent news in the space

Note:

• Growing/shrinking market
• Recent funding
• Acquisitions
• Regulatory changes

Step 6: Synthesize findings

Create a research summary:

## Research

**Researched:** YYYY-MM-DD

### Competitors

| Name         | URL | Pricing | Notes       |
| ------------ | --- | ------- | ----------- |
| Competitor 1 | url | $X/mo   | Brief notes |
| Competitor 2 | url | Free    | Brief notes |

### Discussion Insights

- Insight 1 (source: Reddit thread)
- Insight 2 (source: HN discussion)
- Pain point people express
- What's missing from current solutions

### Market Signals

- Signal 1
- Signal 2

### Assessment

Brief assessment:

- Is the problem real?
- How crowded is the market?
- Potential differentiation?
- Red flags?

### Next Steps

- [ ] Action 1
- [ ] Action 2

Step 7: Update the note

Append the research summary to the note.

Update frontmatter:

updated: YYYY-MM-DD

If the note is a project/idea, optionally add:

status: active

Step 8: Report completion

Research complete for "voice-notes-app"

Found:
- 5 competitors (2 direct, 3 adjacent)
- 3 relevant Reddit discussions
- 1 HN thread with 200+ comments

Key insight: Users frustrated with transcription accuracy.
Gap identified: No solution focuses on [specific niche].

Research appended to voice-notes-app.md

Research Depth

Focus on breadth over depth:

• Find competitors (don't deep-dive each)
• Skim discussions for patterns
• Note market signals
• Keep it actionable

Target: 5-10 minutes per note, not hours.

Error Handling

If searches return poor results:

• Note which searches didn't yield useful info
• Suggest alternative search terms
• Ask if user wants to refine the note

If the note is too vague:

• Ask clarifying questions before proceeding
• Don't waste searches on ambiguous queries