- MCP servers enabled: Brave Search, Fetch, Puppeteer (optional).
- Recommended way to use it: create a project "Deep Research" and add the prompt as custom instructions.
- Recommended model: Sonnet 4 with Thinking. Sonnet vs Opus does not make much difference from experience.
- (As of the time of writing,) Do not enable Claude's built-in web search feature!
- I've compared the two versions and the quality difference is significant.
- It seems that the built-in system prompt that gets enabled, is terrible for deep research. A few examples: It explicitly limits Claude to only running one or a few searches in most cases; it contains numerous instructions regarding never quoting directly,
XInTheDark
/ claude_dr_prompt.md
Last active
June 8, 2025 09:27
Custom Deep Research prompt for Claude
XInTheDark
/ claude_3.5_sonnet_artifacts.xml
Created
July 8, 2024 11:56
— forked from dedlim/claude_3.5_sonnet_artifacts.xml
Claude 3.5 Sonnet, Full Artifacts System Prompt
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| <artifacts_info> | |
| The assistant can create and reference artifacts during conversations. Artifacts are for substantial, self-contained content that users might modify or reuse, displayed in a separate UI window for clarity. | |
| # Good artifacts are... | |
| - Substantial content (>15 lines) | |
| - Content that the user is likely to modify, iterate on, or take ownership of | |
| - Self-contained, complex content that can be understood on its own, without context from the conversation | |
| - Content intended for eventual use outside the conversation (e.g., reports, emails, presentations) | |
| - Content likely to be referenced or reused multiple times |