DSPy-Programming-not-prompting-LMs-skills

Review DSPy code for correctness and best practices. Use when you want a code review of your DSPy program, need to check if your AI code follows best practices, want to find anti-patterns in your DSPy usage, or need a quality audit of your AI implementation. Also use for DSPy code review, is my DSPy code correct, review my AI code, best practices check, DSPy anti-patterns, code quality audit, am I using DSPy right, sanity check my AI code, peer review my DSPy program, does this follow DSPy conventions.

Verify and validate AI output before it reaches users. Use when you need guardrails, output validation, safety checks, content filtering, fact-checking AI responses, catching hallucinations, preventing bad outputs, or quality gates. Also used for - AI output looks right but is wrong, how to validate JSON from LLM, LLM returns invalid data, catch bad AI outputs before users see them, output quality gate, AI guardrails for production, verify LLM did not hallucinate fields, post-processing LLM responses. Uses dspy.Refine (iterative with feedback) and dspy.BestOfN (sampling, pick best).

Pick the right DSPy module and architecture for your AI feature. Use when you are not sure whether to use Predict, ChainOfThought, ReAct, or a pipeline, need to choose between DSPy patterns, want architecture advice for your AI feature, or are deciding between a single module and a multi-step pipeline. Also use for which DSPy module should I use, Predict vs ChainOfThought, when to use ReAct, single module vs pipeline, DSPy architecture decision, CoT vs PoT vs ReAct, do I need a pipeline, module selection guide, DSPy pattern selection, how to structure my DSPy program.

Normalize and fix messy data fields using AI. Use when normalizing addresses, standardizing company names, fixing inconsistent date formats, cleaning CSV data before import, correcting typos in bulk data, normalizing phone number formats, standardizing job titles, cleaning up free-text fields, data quality improvement with AI, fixing formatting inconsistencies, bulk data normalization, preparing messy data for analysis, AI-powered data wrangling.

Reduce your AI API bill. Use when AI costs are too high, API calls are too expensive, you want to use cheaper models, optimize token usage, reduce LLM spending, route easy questions to cheap models, or make your AI feature more cost-effective. Also used for GPT-4 costs too much for production, AI bill keeps growing, how to reduce OpenAI costs, optimize LLM token usage, smart model routing saves money, prompt is too long and expensive, cheaper than GPT-4 with same quality.

Describe your AI problem and get routed to the right skill with a ready-to-use prompt. Use when you are not sure which ai- skill to use, want help picking the right approach, or just want to describe what you need in plain language. Also use this when someone says I want to build an AI that..., how do I make my AI..., or describes any AI/LLM task without naming a specific skill, I need AI but do not know where to start, which AI pattern should I use, what is the best way to add AI to my app, recommend an AI approach, AI feature discovery, too many AI options, overwhelmed by AI frameworks, just tell me what to build, new to DSPy, beginner AI project help, which LLM pattern fits my use case, confused about AI architecture, help me figure out my AI approach.

Fine-tune models on your data to maximize quality and cut costs. Use when prompt optimization hit a ceiling, you need domain specialization, you want cheaper models to match expensive ones, you heard fine-tuning will make us AI-native, you have 500+ training examples, or you need to train on proprietary data. Also use when you have spent weeks of manual iteration with no systematic improvement path, or manual prompt tuning got you to a working system but quality plateaued. Covers DSPy BootstrapFinetune, BetterTogether, model distillation, and when to fine-tune vs optimize prompts, LoRA vs full fine-tune, when to fine-tune vs few-shot, distill GPT-4 into a smaller model, teacher-student model training, custom model training with DSPy, model distillation, make a cheap model as good as GPT-4.

Measure and improve how well your AI works. Use when AI gives wrong answers, accuracy is bad, responses are unreliable, you need to test AI quality, evaluate your AI, write metrics, benchmark performance, optimize prompts, improve results, or systematically make your AI better. Also used for spent hours tweaking prompts, trial and error prompt engineering is not working, quality plateaued early, stale prompts everywhere in your codebase, my AI is only 60% accurate, how to measure AI quality, AI evaluation framework, benchmark my LLM, prompt optimization not working, systematic way to improve AI, AI accuracy plateaued, DSPy optimizer tutorial, MIPROv2 optimization, how to go from 70% to 90% accuracy.

Scaffold a new AI feature powered by DSPy. Use when adding AI to your app, starting a new AI project, building an AI-powered feature, setting up a DSPy program from scratch, or bootstrapping an LLM-powered backend. Also used for DSPy quickstart, DSPy hello world, first DSPy program, getting started with DSPy, new to AI development, add AI to existing Python app, AI feature from zero to working, scaffold AI project structure, best practices for AI project setup, where do I even begin with LLMs, AI boilerplate code, starter template for AI features, bootstrap AI backend, simple AI project template, how to structure an AI codebase, AI mvp in a day, proof of concept AI feature, DSPy project structure best practices.

Make your AI give the same answer every time. Use when AI gives different answers to the same question, outputs are unpredictable, responses vary between runs, you need deterministic AI behavior, or your AI is unreliable. Also used for same input gives different output every time, prompt sensitivity causes output changes with minor wording tweaks, reordering examples shifts accuracy dramatically, same prompt gives different results every run, AI is non-deterministic, need reproducible AI results, LLM output keeps changing, how to make LLM deterministic, consistent JSON from LLM, reduce output variance, AI flaky in production, stable AI outputs for production.

Find and merge duplicate records across datasets using AI. Use when deduplicating contacts, merging customer records, entity resolution, matching records across systems, finding duplicate tickets, fuzzy matching company names, CRM deduplication, record linkage across databases, matching people across data sources, identifying the same entity in different formats, contact merge, account deduplication, consolidating duplicate entries.

Auto-moderate what users post on your platform. Use when you need content moderation, flag harmful comments, detect spam, filter hate speech, catch NSFW content, block harassment, moderate user-generated content, review community posts, filter marketplace listings, or route bad content to human reviewers. Also used for build content moderation system, UGC moderation at scale, user-generated content filter, trust and safety tooling, hate speech detection model, NSFW detection API, toxic comment classifier, automated abuse detection, report and flag system with AI, content policy enforcement, marketplace listing moderation, DSPy classification with severity scoring, confidence-based routing, reward-based policy enforcement.

Pull structured data from messy text using AI. Use when parsing invoices, extracting fields from emails, scraping entities from articles, converting unstructured text to JSON, extracting contact info, parsing resumes, reading forms, pulling data from transcripts (VTT, LiveKit, Recall), extracting fields from Langfuse traces, or any task where messy text goes in and clean structured data comes out. Also use when emails are messy and lack structure, or structured data extraction from unstructured content is unreliable., extract entities from text, parse PDF with AI, structured extraction from unstructured text, OCR plus AI extraction, convert email to structured data, pull fields from documents automatically, AI data entry automation, invoice parsing, resume parsing with AI, medical record extraction.

Plan a multi-phase AI feature before building it. Use when you have a PRD or project idea and need to figure out the execution order, which skills to use in what sequence, or how to break an ambitious AI project into phases. Also use when you want to scope an AI feature, create a phased rollout plan, or figure out dependencies between AI components., help me figure out how to execute this, plan my AI feature, what order should I build this in, AI project roadmap, break this into phases, scope an AI feature, phased AI rollout, AI feature planning, multi-phase AI project, AI project dependencies, which skills do I need, AI execution plan

Build AI that answers questions about your database. Use when you need text-to-SQL, natural language database queries, a data assistant for non-technical users, AI-powered analytics, plain English database search, or a chatbot that talks to your database. Covers DSPy pipelines for schema understanding, SQL generation, validation, and result interpretation., text-to-SQL that actually works, AI SQL generation is unreliable, let non-technical users query data, build a data analyst chatbot, business intelligence with AI, self-service analytics, AI dashboard queries, ask questions about my database in English, SQL copilot, AI-powered data exploration, Metabase alternative with AI, chat with your Postgres, natural language analytics, data chatbot for stakeholders.

Make AI solve hard problems that need planning and multi-step thinking. Use when your AI fails on complex questions, needs to break down problems, requires multi-step logic, needs to plan before acting, gives wrong answers on math or analysis tasks, or when a simple prompt is not enough for the reasoning required. Covers ChainOfThought, ProgramOfThought, MultiChainComparison, and Self-Discovery reasoning patterns in DSPy., AI gives shallow answers, LLM does not think before answering, chain of thought prompting, make AI show its work, AI fails at math, complex analysis with LLM, multi-step problem solving, AI reasoning errors, LLM logic mistakes, think step by step DSPy, AI cannot do basic arithmetic, deep reasoning with language models, self-consistency for better answers, tree of thought.

Build AI that recommends products, articles, or content based on user preferences. Use when building product recommendations, you-might-also-like features, personalizing feeds, content discovery, ranking by user preference, suggesting related items, building a recommendation engine, collaborative filtering with LLMs, re-ranking search results by relevance, curating personalized playlists, next-best-action suggestions, upsell recommendations, similar item matching, AI-powered content curation.

Strip PII and sensitive data from text before processing with AI. Use when redacting personal information, GDPR compliance, anonymizing customer data, masking credit cards, redacting PHI for HIPAA, stripping emails and phone numbers, de-identifying medical records, removing names from transcripts, PII detection and replacement, building a data anonymization pipeline, sanitizing text before sending to LLMs, pre-processing sensitive documents, privacy-preserving AI pipelines.

Request or contribute a new AI skill that does not exist yet. Use when DSPy supports something but there is no skill for it — helps you build the skill and submit a PR, or file an issue requesting it. Also use when the user says there should be a skill for this, can we make a skill, I want to contribute a skill, none of the ai- skills cover my use case, how do I add a new skill, submit a skill, or open a PR for a missing skill.

Rewrite text to match a different tone, reading level, or audience using AI. Use when rewriting content in a different tone, simplifying legal language, adapting text for a different audience, converting technical docs to plain English, making formal text casual, adjusting reading level, matching brand voice in existing content, paraphrasing for clarity, converting jargon-heavy text to simple language, tone transformation, style transfer for text, rewriting marketing copy, making content more accessible, executive summary from technical report.

Put your AI behind an API. Use when you need to serve AI features as web endpoints, add AI to an existing backend, deploy AI for other services to call, wrap a DSPy program in REST or HTTP, build an AI microservice, or put a language model behind FastAPI or Flask. Also use for deploy AI model to production, AI REST API, serve DSPy program over HTTP, Docker AI service, AI endpoint for mobile app, how to productionize my AI, LLM behind a web API, AI microservice architecture, AI backend for React app, put my AI in production, AI API for frontend to call.

Stop your AI from making things up. Use when your AI hallucinates, fabricates facts, is not grounded in real data, does not cite sources, makes unsupported claims, or you need to verify AI responses against source material. Also use when your LLM makes up facts, responses are disconnected from the input, or outputs are not grounded in source documents. Covers citation enforcement, faithfulness verification, grounding via retrieval, confidence thresholds, and evaluation of anti-hallucination quality. Also used for AI makes up citations, LLM fabricates data, ground AI in source documents, RAG but AI still hallucinates, force AI to cite sources, factual accuracy for AI, prevent AI from inventing facts, AI confident but wrong, LLM confabulation, hallucination detection, verify AI claims against documents.

Condense long content into short summaries using AI. Use when summarizing meeting notes, condensing articles, creating executive briefs, extracting action items, generating TL;DRs, creating digests from long threads, summarizing customer conversations, or turning lengthy documents into bullet points. Also used for AI summary too generic, summarize Slack threads, condense customer feedback, meeting transcript summary, executive summary generator, AI-powered digest, summarize legal documents, TLDR for long emails, abstractive summarization, extractive summary with AI, bullet point summary from long text, summarize research papers, call transcript summary, weekly digest generator, summarize support tickets, AI loses important details when summarizing, key takeaways extraction.

Build AI that takes actions, calls APIs, and does things autonomously. Use when you need AI to call APIs, use tools, perform calculations, search the web and act on results, interact with databases, or do multi-step tasks. Also AI that does things not just talks, tool-using AI agent, AI calls external APIs, function calling with DSPy, build AI that books appointments, AI workflow automation, agent that searches and acts on results, AI that updates databases, autonomous AI agent, AI performs multi-step tasks, give LLM access to tools, agentic AI workflow, AI agent for DevOps, build AI assistant that takes actions, MCP tool integration with AI, AI that can browse and click, LLM with tool access.

See exactly what your AI did on a specific request. Use when you need to debug a wrong answer, trace a specific AI request, profile slow AI pipelines, find which step failed, inspect LM calls, view token usage per request, build audit trails, or understand why a customer got a bad response. Covers DSPy inspection, per-step tracing, OpenTelemetry instrumentation, and trace viewer setup., debug slow AI response, why is my AI pipeline slow, trace LLM token usage, OpenTelemetry for AI, Langfuse tracing, AI observability per request, debug wrong AI answer for specific user, which LLM call failed, latency profiling for AI, audit trail for AI decisions, inspect what the AI actually saw, per-request AI debugging, production AI request logs, DSPy inspect_history, trace AI reasoning steps.

Translate text between languages with AI while preserving brand voice and terminology. Use when translating app copy to Spanish, localizing marketing content, multilingual support tickets, i18n with AI, machine translation with brand voice, translating product descriptions, localizing help docs, batch translating i18n JSON files, glossary-enforced translation, AI-powered localization pipeline, translate content without losing tone, bilingual customer support, auto-translate user-facing strings, locale-aware content generation.

Analyze images and extract structured data from visual content using AI. Use when analyzing product photos, extracting text from screenshots, generating alt text for accessibility, visual question answering, categorizing images by content, reading receipts and invoices from photos, OCR with AI, describing images for search indexing, product photo categorization, document image processing, chart and graph extraction, UI screenshot analysis, image-to-structured-data pipelines.

Use when you have already tried prompt-only optimization and want the next level — jointly tuning prompts and model weights for maximum quality. Common scenarios - you have maxed out prompt optimization and need the next level, combining instruction tuning with weight tuning for maximum quality, making a small model match a large model through joint optimization, or squeezing the last few percent of accuracy. Related - ai-fine-tuning, ai-improving-accuracy, ai-cutting-costs. Also used for dspy.BetterTogether, joint prompt and weight optimization, beyond prompt engineering, combine fine-tuning with prompt optimization, maximum possible quality from DSPy, hybrid optimization strategy, prompt optimization hit a ceiling, fine-tune and optimize prompts at the same time, advanced DSPy optimization, best possible accuracy, what to try after MIPROv2, next level AI quality.

Use when you need structured source attribution in AI responses — verifiable citations that link claims to specific passages in source documents. Common scenarios - RAG with citation extraction, grounded answers with document references, legal or compliance use cases requiring source proof, building AI that cites its sources, or verifying which document an answer came from. Related - ai-stopping-hallucinations, ai-searching-docs, dspy-retrieval. Also used for dspy.experimental.Citations, dspy.experimental.Document, cite sources in DSPy, structured citations, source attribution, verify which document answer came from, RAG with citations, grounded answers with references, citation extraction, Anthropic Citations API DSPy, cited_text, document_index, adapt_to_native_lm_feature, parse_lm_response citations, streaming citations.

Use when you need to prepare training/dev data for DSPy optimizers — loading from CSV/JSON/HuggingFace, creating Examples, setting input keys, or building train/dev splits. Common scenarios - loading a CSV of labeled examples for optimization, converting HuggingFace datasets to DSPy format, creating train/dev/test splits, building Examples with proper input keys, converting JSON data for DSPy, or preparing evaluation datasets. Related - ai-generating-data, dspy-evaluate. Also used for dspy.Example, dspy.Dataset, load training data for DSPy, CSV to DSPy examples, HuggingFace dataset in DSPy, prepare data for optimization, input_keys in DSPy, train dev split for DSPy, how to format data for DSPy optimizer, labeled examples format, create examples from JSON, what format does DSPy expect, dataset preparation for DSPy, with_inputs in DSPy Example, build evaluation dataset.

Use when you have multiple optimized versions of a program and want to combine them — voting, averaging, or routing across program variants for more robust outputs. Common scenarios - you have optimized several versions of a program and want to combine the best ones, using majority voting across multiple programs for higher accuracy, building a robust system by routing to different specialized programs, or reducing variance by averaging outputs. Related - ai-improving-accuracy, ai-making-consistent, dspy-bootstrap-rs. Also used for dspy.Ensemble, combine multiple optimized programs, majority voting across models, ensemble of DSPy programs, voting for reliability, reduce variance with multiple programs, aggregate predictions, combine outputs from different optimizers, when one program is not reliable enough, model committee, ensemble for production robustness, multiple programs one answer.

Use when you need to measure how well your DSPy program performs — writing metrics, scoring against a dev set, or comparing before/after optimization. Common scenarios - measuring accuracy before and after optimization, writing custom metrics for your task, scoring a program against a held-out dev set, comparing two prompt strategies, building a test suite for AI quality, or running regression tests on AI outputs. Related - ai-improving-accuracy, ai-scoring, ai-monitoring. Also used for dspy.Evaluate, dspy.evaluate, write DSPy metric function, measure AI accuracy, evaluate DSPy program, dev set evaluation, before and after optimization comparison, custom scoring function, test AI quality systematically, AI regression testing, metric-driven development, how to know if my DSPy program improved, score predictions against labels, evaluation harness for LLM, CI/CD for AI quality.

LLM observability for DSPy with Langfuse -- auto-trace every LM call, attach scores and evaluations, run annotation queues for human review, and track experiments across prompt versions. Use when you want to set up Langfuse, langfuse.com, openinference-instrumentation-dspy, trace DSPy calls, LLM observability with scores, annotation queues, or experiment tracking. Also used for langfuse setup, pip install langfuse, DSPy trace viewer, langfuse vs phoenix, langfuse vs langtrace, observe decorator with DSPy, self-hosted tracing with evaluation, production LLM monitoring with scoring.

Use when you need to configure which language model DSPy uses — setting up providers, API keys, model parameters, or assigning different models to different pipeline stages. Common scenarios - setting up OpenAI or Anthropic API keys, configuring model parameters like temperature and max_tokens, using different models for different pipeline stages, switching between providers, using local models with Ollama or vLLM, or setting up Azure OpenAI. Related - ai-switching-models, ai-cutting-costs, ai-kickoff. Also used for dspy.LM, dspy.configure, configure language model in DSPy, OpenAI API key setup DSPy, Anthropic Claude with DSPy, use Ollama with DSPy, local model DSPy, Azure OpenAI DSPy setup, model temperature and max_tokens, different models per module, multi-model DSPy pipeline, vLLM with DSPy, change provider without changing code, model configuration DSPy.

Use when you want the highest-quality prompt optimization DSPy offers — jointly optimizes instructions and few-shot demos, with auto=light/medium/heavy presets. Common scenarios - you want the best possible accuracy from prompt optimization, jointly tuning instructions and few-shot demonstrations, using auto presets for different compute budgets, or when COPRO or BootstrapFewShot alone are not reaching your accuracy target. Related - ai-improving-accuracy, dspy-copro, dspy-bootstrap-few-shot. Also used for dspy.MIPROv2, best DSPy optimizer, highest quality optimization, auto=light medium heavy, joint instruction and demo optimization, most powerful prompt optimizer, MIPROv2 vs COPRO vs BootstrapFewShot, which optimizer should I use, state of the art prompt optimization, when to use MIPROv2, optimize both instructions and examples, heavy optimization for production, best optimizer for accuracy.

Use when you want higher accuracy by generating multiple reasoning chains and selecting the best answer — trading speed for quality on critical outputs. Common scenarios - high-stakes decisions where you want multiple reasoning paths compared, classification tasks where one chain of thought is not reliable enough, improving accuracy by generating several answers and selecting the best-reasoned one, or tasks where different reasoning approaches yield different answers. Related - ai-reasoning, ai-improving-accuracy, dspy-chain-of-thought. Also used for dspy.MultiChainComparison, compare multiple reasoning chains, select best reasoning path, multi-path reasoning, vote across chain-of-thought outputs, more reliable than single CoT, deliberation for hard problems, when one reasoning chain is not enough, robust reasoning through comparison, ensemble reasoning, trade speed for accuracy on critical tasks.

Use when the mapping from input to output is straightforward and does not need reasoning steps — simple classification, extraction, formatting, or Q&A where minimal latency matters. Common scenarios - simple classification tasks, basic extraction, format conversion, straightforward Q&A, or any task that does not benefit from chain-of-thought reasoning — when you want the fastest possible LM call. Related - ai-sorting, ai-parsing-data, dspy-chain-of-thought. Also used for dspy.Predict, simplest DSPy module, basic LM call in DSPy, direct prediction no reasoning, when to use Predict vs ChainOfThought, fast classification with DSPy, minimal latency LM call, simple input-output mapping, Predict vs ChainOfThought, zero overhead DSPy call, straightforward text generation, quick extraction without reasoning, one-shot prediction, basic DSPy hello world.

DSPy typed wrappers (dspy.Image, dspy.Audio, dspy.Code, dspy.History, dspy.File, dspy.Reasoning, dspy.Tool, dspy.ToolCalls) for multimodal data, files, and structured outputs in signatures. Use when working with non-text inputs like images, audio, or code, passing PDFs or documents to the LM, capturing native reasoning traces from reasoning models, building multimodal AI pipelines, processing images alongside text, handling audio transcription inputs, working with code files as typed inputs, or managing conversation history in multi-turn chatbots. Also used for multimodal DSPy, image input in DSPy signature, process images with DSPy, audio input in DSPy, dspy.File, pass PDF to language model, document input in DSPy, dspy.Reasoning, capture thinking traces, native reasoning output, dspy.Tool, dspy.ToolCalls, typed fields in signatures, non-text data in DSPy, vision model with DSPy, Claude vision with DSPy, multimodal pipeline, image classification with DSPy, pass images to language model, conversation history

Use Qdrant as a vector database with DSPy, or connect any vector DB (Pinecone, ChromaDB, Weaviate) with custom retrievers. Use when you want to set up Qdrant, QdrantRM, dspy-qdrant, vector database for DSPy, vector search, hybrid search, or build custom retrievers for Pinecone, ChromaDB, or Weaviate. Also used for qdrant, dspy-qdrant, QdrantRM, vector database, vector search, pinecone DSPy, chromadb DSPy, weaviate DSPy, vector DB for DSPy, pip install dspy-qdrant, qdrant docker, qdrant cloud, hybrid search DSPy, sparse dense vectors, custom dspy.Retrieve, which vector DB for DSPy, DSPy 3.0 retriever removed.

Use Ragas to evaluate DSPy RAG pipelines with decomposed metrics. Use when you want to evaluate RAG quality, measure faithfulness, context precision, context recall, answer relevancy, or diagnose retriever vs generator issues. Also used for ragas, pip install ragas, ragas evaluate, RAG evaluation, faithfulness metric, context precision, context recall, answer relevancy, answer correctness, decomposed RAG metrics, ragas dspy, DSPyOptimizer ragas, ragas[dspy], EvaluationDataset, ragas vs dspy.Evaluate, which RAG metric, retriever vs generator quality.