AI Coding Session Auto-Memory and Context Injection Template

You are an expert in designing memory systems for AI coding assistants. Your task is to create a comprehensive session memory and context injection framework. ## Requirements Design a system that: 1. **Auto-captures** key decisions, file changes, architecture patterns, and error resolutions during coding sessions 2. **Compresses** raw session logs into structured, retrievable knowledge using hierarchical summarization 3. **Injects** relevant context into future sessions based on semantic similarity and recency 4. **Manages** memory lifecycle: capture → compress → index → retrieve → inject → prune ## Output Format Provide: - Memory schema (JSON) with fields: timestamp, category, content, embeddings_key, relevance_score - Capture rules: what to save (decisions, errors, patterns) vs skip (routine operations) - Compression pipeline: raw logs → structured summaries → knowledge atoms - Retrieval strategy: when and how to inject context (session start, on-demand, triggered) - Pruning policy: TTL, relevance decay, deduplication ## Context Target coding agent: [AGENT_NAME] Project type: [PROJECT_TYPE] Typical session length: [DURATION] Key pain points: [PAIN_POINTS] Start by analyzing the agent's memory needs, then output the complete framework with implementation examples.