304 prompts available in the DeepSeek directory
Ready-to-use prompt templates for common tasks. Each prompt is tested and optimized — copy it, fill in your details, and get better results from the AI.
POC: Fully autonomous weapons systems via LLMs — deepseek-v4-pro reasoning model with unrestricted system prompt demonstrates autonomous offensive cyber operations capability for classification analysis and regulatory discussion. ECCN: 4D004 / EAR99.
Codex Cost Router Skills - preflight task routing skills for Codex to reduce token waste, choose model tiers, reasoning effort, permissions, and execution prompts.
Test and compare LLM prompts across providers — automated quality scoring, latency benchmarks, cost analysis. YAML-driven, works with any OpenAI-compatible API.
基于 FastAPI 的 DeepSeek Chat 反向代理,将 DeepSeek 网页版的 API 转换为 OpenAI 兼容格式。 支持流式/非流式对话、专家模式、深度思考(reasoning_content)、工具调用(DSML prompt injection)。 自动处理 PoW 鉴权挑战,无需官方 API Key。
Frozen reasoning LLM (DeepSeek-V4-Flash) as an in-context surrogate for lithium-electrolyte ionic conductivity on CALiSol-23 — selector ablation, chain-of-thought pattern mining, and a one-sentence prompt change that cut worst-case error 87×.
Local DeepSeek V4 proxy: reduces hallucinations strengthens system prompt following stops reasoning loops repairs empty tool-call replies — with built-in session analysis and GUI. / 本地 DeepSeek V4 代理:降低幻觉、增强 system prompt 遵循、防思维循环、修复空回复 自带会话分析与 GUI。
Leverage DeepSeek Coder for complete, production-quality code implementations. Excels at algorithmic problems, system design, and competitive programming.
Structured reasoning framework leveraging DeepSeek R1's advanced thinking capabilities. For complex math, logic, coding, and multi-step analysis problems.
Head-to-head reasoning benchmark across Opus 4.7, GPT-5.5, and DeepSeek V4 Pro on hard discriminating prompts. Independent third-party judge from a vendor outside the contestant set, anonymized A/B grading, and full per-task scoring tables exported as JSON and SVG.
Interpret legal contract clauses step by step with precedent references, statutory analysis, and practical implications.
Perform detailed financial analysis including ratio analysis, trend identification, and comparative benchmarking.
Solve linear algebra problems with step-by-step matrix operations, eigenvalue computation, and vector space analysis.
Given incomplete evidence, use DeepSeek R1 to generate the most likely explanations through abductive reasoning.
Full-featured feature flag system with targeting, percentage rollouts, and React hooks.
Comprehensive web performance analysis targeting LCP, CLS, INP, and bundle size with code fixes.
Generates a complete GitHub Actions CI/CD pipeline with testing, security scanning, and deployment.
Comprehensive security audit prompt covering OWASP Top 10 with severity ratings and fixes.
Complete WebSocket notification system with rooms, reconnection, and React hooks.
Complete Docker Compose setup with database, cache, storage, and development tooling.
Multi-step form wizard with per-step validation, auto-save drafts, and animated transitions.
Auto-generates a comprehensive README by analyzing your actual codebase structure and dependencies.
Full observability setup with structured logging, distributed tracing, metrics, and alerting.
Design microservices architectures with DeepSeek V3 using DDD, event-driven patterns, and comprehensive observability with migration planning.
Guide students to understanding through the Socratic method using DeepSeek R1, asking probing questions instead of giving direct answers.