
Deep Research Methodology
Makes the agent run multi-angle web research before answering or generating content
- Repository
- bytedance/deer-flow
- Stars
- 84,876
- Licence
- MIT
- Files
- 1 file
What it does
When installed, this skill tells the agent to stop relying on a single web search or its general memory. Instead it follows a structured process: survey the topic broadly, break it into subtopics, run targeted searches, read key sources in full, and check for data, examples, expert views, and criticisms before writing an answer or content. It also tells the agent to use the current date when searching for recent information.
Great for
- People who want well-sourced answers to research questions rather than quick summaries
- Anyone preparing a report, presentation, article, or UI design that needs current facts
- Users who want the agent to look at both benefits and limitations of a topic
Skip it if
- Quick factual questions where one lookup is enough; the process is slow and token-heavy by design
- Situations needing verified accuracy for medical, legal or financial decisions, since the skill only guides searching and does not check sources
- Offline or privacy-sensitive work, because it depends on web searches and page fetches that send queries to external services
Try asking
- “Research the current state of AI in healthcare diagnostics and summarise the main challenges”
- “Compare two cloud databases and find recent expert commentary on each before recommending one”
- “Gather statistics and case studies before I write a slide deck on remote work trends”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the MIT licence and an attribution note.
deep-research.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip deep-research.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip deep-research.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- State the current date and your purpose (for example, a report for a specific audience) so the agent picks the right search precision and sources.
- Ask the agent to list the sources it used so you can check the key claims yourself.
- For time-sensitive topics, say whether you mean today, this week, or this year so the date-based searches match your need.
What's inside
- SKILL.md
Redistributed unmodified from bytedance/deer-flow under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.