LLM Prompt Library
19 production-ready prompt templates across 7 categories. Each template includes token estimates and real cost calculations so you know exactly what you'll pay before deploying.
Copy the prompts as-is or adapt to your use case. Replace {VARIABLE} placeholders with your actual values.
How to use these prompts
19 prompts found
Customer Inquiry Response
beginnerDraft a professional, empathetic reply to a customer support message. Adapt tone based on the urgency and sentiment of the inquiry.
{COMPANY_NAME}{CUSTOMER_MESSAGE}{PRODUCT_CONTEXT}Support Ticket Triage & Priority
beginnerClassify incoming support tickets by category, urgency, and required team. Returns structured JSON for routing automation.
{SUBJECT}{TICKET_BODY}FAQ Answer from Knowledge Base
intermediateAnswer customer questions grounded strictly in a provided knowledge base. Refuses to speculate beyond the given context.
{SUPPORT_EMAIL}{KNOWLEDGE_BASE}{QUESTION}Escalation Risk Detection
intermediateDetect whether a customer conversation is at risk of churn or requires manager escalation. Returns risk score and reason.
{CONVERSATION_HISTORY}{ACCOUNT_TIER}Code Review with Actionable Feedback
intermediatePerform a structured code review covering correctness, security, performance, and readability. Returns prioritized issues with fix suggestions.
{LANGUAGE}{CONTEXT_DESCRIPTION}{CODE}Bug Fix with Explanation
intermediateIdentify the root cause of a bug, explain it clearly, and provide the corrected code with a before/after diff.
{LANGUAGE}{ERROR_DESCRIPTION}{CODE}Docstring & API Documentation Generator
beginnerGenerate comprehensive docstrings for functions, classes, or modules. Covers parameters, return values, exceptions, and usage examples.
{LANGUAGE}{STYLE}{CODE}Code Translation Between Languages
intermediateTranslate code from one programming language to another, preserving logic, idioms, and performance characteristics of the target language.
{SOURCE_LANGUAGE}{TARGET_LANGUAGE}{CODE}Structured Data Extraction to JSON
beginnerExtract specific fields from unstructured text into a defined JSON schema. Handles missing fields gracefully with null values.
{JSON_SCHEMA}{SOURCE_TEXT}Named Entity Recognition (NER)
beginnerExtract and classify named entities (people, organizations, locations, dates, amounts) from text. Returns structured JSON with confidence.
{SOURCE_TEXT}Table Extraction from Documents
intermediateExtract tabular data from invoices, reports, or prose descriptions. Returns a clean CSV or JSON array of rows.
{DOCUMENT_TEXT}{EXPECTED_COLUMNS}Executive Summary (Long Document)
beginnerGenerate a structured executive summary of long documents (reports, research papers, contracts). Extracts key findings, decisions, and action items.
{TITLE}{DOCUMENT_TYPE}{DOCUMENT_TEXT}Meeting Notes → Action Items
beginnerConvert raw meeting transcripts or notes into structured summaries with decisions made and clearly assigned action items.
{MEETING_NOTES}Blog Post Outline from Topic
beginnerGenerate a detailed, SEO-optimized blog post outline with H2/H3 structure, key points per section, and suggested word counts.
{TOPIC}{AUDIENCE}{PRIMARY_KEYWORD}{TONE}{WORD_COUNT}Product Description Generator
beginnerWrite compelling e-commerce product descriptions that highlight benefits, address objections, and include relevant keywords.
{WORD_COUNT}{PRODUCT_NAME}{CATEGORY}{FEATURES}{TARGET_CUSTOMER}{PRICE}Sentiment Analysis with Confidence
beginnerClassify text sentiment with confidence score and aspect-level breakdown. Returns structured JSON for downstream processing.
{TEXT}User Intent Classification for Chatbots
intermediateClassify user messages into predefined intents for chatbot routing. Handles multi-intent messages and slot extraction.
{INTENT_LIST}{USER_MESSAGE}{CONTEXT}RAG Question Answering with Citations
intermediateAnswer questions based on retrieved document chunks. Provides citations, handles unanswerable questions, and never hallucinates.
{RETRIEVED_DOCUMENTS}{QUESTION}Document Relevance Reranking
advancedRerank a list of retrieved document chunks by relevance to a query. More accurate than vector similarity alone for complex questions.
{QUERY}{DOCUMENTS}