Reasoning model system prompt for catching inconsistencies in tabular data. Returns a tight JSON contract that downstream code can branch on. Source: https://cookbook.openai.com/ex

Opening section of Devin's leaked system prompt. Establishes Devin's identity as an autonomous SWE, defines when to communicate with the user, and lays out the planning vs standard

Also from the cookbook summarization guide. The 'reduce' step that combines per chunk summaries generated cheaply with Haiku into one coherent, sectioned final summary using Sonnet

Microsoft's documented chatbot template. The two line instruction block is what keeps the assistant from answering off topic questions, and the IRS fallback URL prevents hallucinat

Dave Hulbert's plain English distillation of Tree of Thoughts. Works on any chat model no search algorithm required and dramatically improves multi step reasoning. Source: https://

A Claude skill that creates three structured spec files requirements, design, and tasks before any code is written, giving AI agents a shared source of truth. Use it when starting

planningchain-of-thoughttemplatesystem-promptagentClaude 3.7 SonnetClaude Sonnet 4.6Claude Opus 4.7

Prompt Curator120repasked 6 days ago

Three short agent prompts from the Swarm style orchestration cookbook. Each agent's "routine" is what the LLM treats as its operating procedure when control is handed off. Source:

Opening sections of Perplexity's leaked system prompt. Strict on inline numeric citations, markdown structure, no moralization, and no "based on search results" hedging. Source: ht

From the basic workflows.ipynb notebook in patterns/agents . Demonstrates the 'route' workflow: classify a ticket, then dispatch to a specialized sub prompt. Reproduced verbatim ar

Verbatim system prompt from the Anthropic Research multi agent pattern. The 'research lead' classifies a user query as depth first, breadth first, or straightforward, then plans an

Entity extraction system prompt that drives a tool calling search over a product database. Notice how the prompt enumerates valid categories so the model snaps its parameter values

The community standard SDXL negative prompt. Weighted parentheses e.g. :1.4 push the model harder against the lowest quality artifacts while leaving room for style. Source: https:/

Chapter 4 Separating Data and Instructions . Compares the un tagged form 'Yo Claude. {EMAIL} < Make this email more polite' with the safer tagged form below. The XML wrap is what s

From how to make sql queries.ipynb in the cookbook. A minimal, schema aware text to SQL prompt that constrains the model to output only the SQL. A more reliable extended version, f

A high detail SDXL prompt that fuses art history references Max Ernst, Yinka Shonibare, shin hanga with photographic specs. Pair with the negative prompt to keep the render clean.

Classic Wei et al. few shot CoT exemplars. Each example shows the reasoning trace explicitly, which conditions the model to do the same on the final query. Source: https://www.prom

Wang et al. self consistency uses the same CoT shape as Wei et al., then samples multiple reasoning paths and majority votes. Use these exemplars verbatim and sample at temperature

From the cookbook's summarization guide. A focused prompt for sublease/legal agreement summarization that uses a system role, explicit section headers, an XML output format, prefil

From the official Anthropic prompt engineering guide for the Claude 4.7 family. Two single line additions that calibrate verbosity in either direction useful as a system prompt fra

Conducts a turn based mock interview for any role. Includes a templated variable so you can swap the position without rewriting the prompt. Source: https://github.com/f/awesome cha