Skill UI

This skill details advanced techniques for enhancing the effective context capacity of LLMs. It outlines four core strategies—Compaction, Observation Masking, KV-Cache Optimization, and Context Partitioning—to prevent context limits from constraining task complexity. By strategically managing token usage, users can reduce operational costs, minimize latency, and implement robust, long-running agent systems for handling large documents or conversations.

Automates inference latency profiler tasks in ML deployment scenarios, offering step-by-step guidance on model serving, MLOps pipelines, monitoring, and production optimization, generating production-ready code and validating outputs against best practices.

This guide provides detailed strategies for optimizing Qdrant's operational performance. It covers critical areas such as search speed (latency and throughput), efficient vector indexing strategies, and advanced memory management techniques. Use this when you need to improve the speed, scalability, and resource efficiency of your vector database deployment.

A comprehensive guide for diagnosing and resolving slow search performance in Qdrant. This resource covers issues related to low QPS (throughput), high latency for single queries, and slow filtered searches. Solutions include tuning HNSW parameters, managing memory pressure, implementing quantization, optimizing indexing, and configuring cluster settings for maximum efficiency.

Context optimization extends effective capacity of limited context windows through strategic compression, masking, caching, and partitioning. The goal is to make better use of available capacity rather than increasing window size. Techniques include compaction, observation masking, KV-cache optimization, and context partitioning to reduce costs and latency in long-running agent systems.

Voice agents enable natural human-AI conversation through speech recognition and synthesis. This skill covers low-latency architectures like speech-to-speech pipelines and STT-LLM-TTS workflows. It includes optimization strategies for sub-800ms response times, voice activity detection, and integration with tools like Deepgram, ElevenLabs, and OpenAI. Ideal for building production-ready voice assistants with natural interaction flows.

Analyze prompt structure to maximize LLM cache reuse, then measure input cost and latency. Covers stable prefixes, user/workspace boundaries, provider-specific cache counters and a cost comparison calculator for OpenAI, Claude and Gemini.