Best Practices for Context Management when Generating Code with AI Agents
Improve AI output quality by carefully managing LLM context.
Inference Concepts
Generated on 15 May 2026
Inference provides a single control plane for managing inference workflows. It includes a Model Catalog where you can view available foundation models, including both DigitalOcean-hosted and third-party commercial models, compare model capabilities and pricing, use routing to match inference requests to the best-fit model, and run inference using serverless or dedicated deployments.
Agent Instructions Best Practices for DigitalOcean Inference
Write effective agent instructions to align responses with your goals, minimize hallucinations, and enhance performance.
Function Instructions Best Practices for DigitalOcean Inference
Write concise and actionable function instructions by defining purpose, specifying outputs, ensuring concise descriptions, and addressing constraints with structured examples.
Model Evaluations Best Practices
Write effective system prompts to maintain a neutral tone, ask focused questions, simplify complexity, and enhance clarity with examples or screenshots.
Prompt Writing Best Practices
Write effective prompts to maintain a neutral tone, ask focused questions, simplify complexity, and enhance clarity with examples or screenshots.
Effectively Group Agents in Workspaces for DigitalOcean Inference
Learn to how you can effectively group agents in workspaces for various purposes.
Chunking Best Practices for DigitalOcean Knowledge Base Indexing
Choose and configure chunking strategies to improve retrieval accuracy, reduce hallucinations, and optimize token usage when indexing knowledge bases.
System Instructions Best Practices for DigitalOcean Knowledge Bases
Write effective system instructions for knowledge base responses by defining the model’s role, limits, communication style, and source requirements.
DigitalOcean Knowledge Base Retrieval Best Practices
Guidance for improving knowledge base retrieval with hybrid search, chunking, filters, and reranking.