research
Deep research via Gemini CLI — runs in background sub-agent so you don't burn your Claude tokens.
Installation
Follow the repository instructions for this skill and your agent. Review the files and dependencies before adding it to your workflow.
View source ↗GuidesSkill instructions
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
--yoloflag auto-approves file operations (non-interactive) - Check
~/clawd/research/for all past research - Always include conversation context in the spawn task for better results
Examples
Read the usage examples in the source repository.
View source ↗Source history
- Indexed on
- 2026-01-20
- Index updated
- 2026-01-30
- Branch
- main
Index and sync dates describe this directory. View the source repository for code history.
GitHub ↗chat Comments (0)
Sign in to join the discussion and leave a comment.
No comments yet. Be the first to share your thoughts!