94 lines
3.2 KiB
Markdown
94 lines
3.2 KiB
Markdown
---
|
|
name: research
|
|
description: Deep research via Gemini CLI — runs in background sub-agent so you don't burn your Claude tokens.
|
|
homepage: https://github.com/google/gemini-cli
|
|
metadata: {"clawdbot":{"emoji":"🔬","requires":{"bins":["gemini"]}}}
|
|
---
|
|
|
|
# Research Skill
|
|
|
|
Conduct deep research on any topic using Gemini CLI via a spawned sub-agent. Uses your Google AI subscription instead of Claude tokens — perfect for long research tasks that would otherwise eat your Clawdbot usage.
|
|
|
|
## How It Works
|
|
|
|
**When user says "Research: [topic]" or asks for deep research:**
|
|
|
|
### Step 1: Clarifying Questions (Always)
|
|
|
|
Before running any research, ask 2-3 quick questions to focus the work:
|
|
|
|
**Start with the goal:**
|
|
> "Before I dive in - what's your goal here? Are you learning about this topic, making a decision, writing something, or just curious?"
|
|
|
|
**Then adapt based on their answer:**
|
|
|
|
If learning/curious:
|
|
- "Any specific aspect you're most interested in?"
|
|
- "How technical should I go? (High-level overview vs deep technical detail)"
|
|
|
|
If decision-making:
|
|
- "What decision are you trying to make?"
|
|
- "Any specific criteria or constraints I should focus on?"
|
|
|
|
If writing/creating:
|
|
- "What's the output? (Blog post, report, presentation?)"
|
|
- "Who's the audience?"
|
|
|
|
**Keep it natural — 2-3 questions max.** Don't interrogate.
|
|
|
|
### Step 2: Spawn Research Agent
|
|
|
|
Once you have context, use `sessions_spawn` to run the research:
|
|
|
|
```
|
|
sessions_spawn(
|
|
task: "Research: [FULL TOPIC WITH CONTEXT]
|
|
|
|
Use Gemini CLI to research this topic. Run:
|
|
|
|
gemini --yolo \"[RESEARCH PROMPT]\"
|
|
|
|
The research prompt should ask Gemini to cover:
|
|
1. Overview & Core Concepts - what is this, terminology, why it matters
|
|
2. Current State - latest developments, major players
|
|
3. Technical Deep Dive - how it works, mechanisms, key techniques
|
|
4. Practical Applications - real-world use cases, tools available
|
|
5. Challenges & Open Problems - technical, ethical, barriers
|
|
6. Future Outlook - trends, predictions, emerging areas
|
|
7. Resources - key papers, researchers, communities, courses
|
|
|
|
Save the output to: ~/clawd/research/[slug]/research.md
|
|
|
|
Be thorough (aim for 500+ lines). Include specific examples and citations.
|
|
|
|
IMPORTANT - When research is complete:
|
|
1. Send a wake event to notify the main agent immediately:
|
|
cron(action: 'wake', text: '🔬 Research complete: [TOPIC]. Key findings: [2-3 bullet points]. Full report: ~/clawd/research/[slug]/research.md', mode: 'now')
|
|
2. When asked to produce an announce message, reply exactly: ANNOUNCE_SKIP",
|
|
label: "research-[slug]"
|
|
)
|
|
```
|
|
|
|
**Important:** Include all context from your conversation in the task so the sub-agent understands the full picture.
|
|
|
|
### Step 3: When You Receive Wake Event
|
|
|
|
You'll receive a wake with the research summary. Then:
|
|
- Share the findings with the user
|
|
- Offer to read the full report or dive deeper on sections
|
|
|
|
## Output Location
|
|
|
|
Research saved to:
|
|
```
|
|
~/clawd/research/<slug>/research.md
|
|
```
|
|
|
|
## Tips
|
|
|
|
- Research typically takes 3-8 minutes depending on complexity
|
|
- Gemini CLI uses your Google AI subscription quota
|
|
- The `--yolo` flag auto-approves file operations (non-interactive)
|
|
- Check `~/clawd/research/` for all past research
|
|
- Always include conversation context in the spawn task for better results
|