Rules
9 rules available in the DeepSeek directory
Development
DeepSeek R1 Code Refactoring Safety Rules
System rules for safe code refactoring with DeepSeek R1, requiring test coverage verification, incremental changes, and behavior preservation checks.
N
Neura MarketWriting
DeepSeek R1 Academic Writing Assistant Rules
System rules for configuring DeepSeek R1 as an academic writing assistant that maintains scholarly tone, proper citations, and disciplinary conventions.
N
Neura MarketOutput Format
DeepSeek R1 Structured Output Enforcer Rules
System rules for forcing DeepSeek R1 to output structured JSON, YAML, or XML responses while still leveraging chain-of-thought reasoning in think tags.
N
Neura MarketArchitecture
DeepSeek Multi-Model Router Rules
Rules for building an intelligent router that selects the optimal DeepSeek model (V3, R1, Coder) based on task type, complexity, and cost constraints.
C
CommunityReasoning
DeepSeek R1 Reasoning Chain Rules
Guidelines for maximizing DeepSeek R1 reasoning quality through proper prompting structure, thinking tag management, and multi-step problem decomposition.
T
Together AI / CommunityConfiguration
DeepSeek R1 Optimal Configuration Guide
Official recommended settings for DeepSeek R1 models including temperature, system prompt handling, and thinking tag enforcement for best performance.
D
DeepSeek OfficialAI/ML
DeepSeek R1 Official System Prompt Guide
Official recommendation: avoid system prompts for R1 — place all instructions in user prompt. For V3, use system prompts to lock in behavior. Temperature 0.6 recommended.
D
DeepSeekEducation
DeepSeek Mathematical Reasoning Configuration
Optimized system configuration for mathematical problem-solving and proof verification with DeepSeek R1.
C
CommunityResearch and Analysis
DeepSeek R1 Research Analyst Configuration
Configuration for using DeepSeek R1 as a research analyst, leveraging its reasoning capabilities for deep analysis and structured research output.
C
Community