AI Memory

Temporal knowledge graph memory system. Use PROACTIVELY at session start to load context, when solving non-trivial problems to check for known solutions, and at session end to persist learnings. Auto-store failure→success learnings when debugging takes 3+ attempts. Auto-store infrastructure context when discovering services/clusters/topology. Triggers on: 'remember this', 'what do you know about', 'save for later', project context questions, approach pivots, or when debugging takes >15 min.

Context recommendation system for reducing action uncertainty. Use PROACTIVELY when user mentions keywords like 'use this kubeconfig', 'our goal is', 'debug this', 'in production', 'deploy to', or when uncertainty about environment, cluster, namespace, or AWS profile drops below 80%.

Experience replay for tool/test/build failures. Invoke IMMEDIATELY when any tool fails (non-zero exit, FAIL/error/panic in output). Looks up known fixes before debugging, records failures and fixes, promotes successful fixes. Triggers on: test failures, build errors, lint failures, compilation errors, any recurring error patterns.