context-manager
Build Intelligent Context Management Systems
This skill helps developers design and implement dynamic context management systems for AI applications, including vector databases, knowledge graphs, and intelligent memory architectures that provide the right information to AI systems at the right time.
Download the skill ZIP
Upload in Claude
Go to Settings → Capabilities → Skills → Upload skill
Toggle on and start using
Test it
Using "context-manager". Design a context management system for a customer support chatbot
Expected outcome:
A comprehensive system design including: (1) Context layers: working memory for active conversation, episodic memory for history, semantic memory for knowledge base. (2) Retrieval strategy: hybrid search combining vector similarity with keyword matching. (3) Context optimization: token budget management, relevance filtering, staleness detection. (4) Agent coordination: handoff protocols, shared context contracts, state synchronization.
Using "context-manager". Optimize RAG performance for 1 million documents
Expected outcome:
Performance optimization strategy covering: (1) Indexing: hierarchical navigable small world (HNSW) indexes with appropriate ef_construction values. (2) Query: hybrid retrieval combining dense embeddings with sparse BM25. (3) Chunking: semantic chunking with 20% overlap for context preservation. (4) Caching: LRU cache for frequent queries, pre-computed embeddings for top queries. (5) Scaling: sharding strategy by document namespace, read replicas for query load.
Security Audit
SafeThis is a prompt-only skill containing only instructional text for AI context engineering. No executable code, network requests, file system access, or command execution patterns detected. Static analysis found 0 files with 0 lines of executable code. Risk score is 0/100 as this skill provides guidance text only, not operational code.
Quality Score
What You Can Build
Enterprise Knowledge Base System
Design a scalable context management system for enterprise document search using vector embeddings and semantic retrieval.
Multi-Agent Customer Support Platform
Create context orchestration for multi-agent customer support with intelligent handoff and state management.
Long-Conversation Memory System
Implement intelligent memory management for sustained AI conversations with episodic and semantic memory layers.
Try These Prompts
Design a context management system for [USE_CASE]. Include components for context assembly, retrieval, and optimization.
Help me implement a vector database solution using [DATABASE_NAME] for [APPLICATION_TYPE]. Include schema design, embedding strategy, and query optimization.
Design a knowledge graph architecture for [DOMAIN] with entity relationships, ontology design, and query optimization strategies.
Create a context handoff protocol for [MULTI_AGENT_SCENARIO] including agent-specific context preparation, state management, and error recovery.
Best Practices
- Apply tiered context strategies: keep critical info in system prompt, use RAG for secondary information, externalize large knowledge bases
- Implement context versioning and change tracking to understand how context evolves over time
- Use hybrid search combining vector similarity with keyword matching for more accurate retrieval
Avoid
- Dumping all available context without prioritization - leads to token limits and reduced relevance
- Ignoring context staleness - serving outdated information degrades user trust
- Over-engineering before understanding actual retrieval needs - start simple, measure, then optimize