context-engineering-advisor

Diagnose context stuffing in RAG workflows by applying adaptive diagnostic questions to determine if context is necessary, retrievable, or harmful, then implement structural fixes like contextual retrieval tuning to reduce noisy tokens.

Context engineering is the deliberate structuring of memory and retrieval to improve attention density, whereas context stuffing blindly overloads windows with unfocused data, degrading model accuracy and increasing token costs.

Design multi-agent memory architecture using a two-layer system separating short-term conversational memory from long-term persistent storage, guided by agent boundary prescriptions to stabilize workflows and increase output consistency.

Inconsistent AI outputs often stem from context stuffing, where large unfocused context windows degrade model performance, requiring a Research-Plan-Reset-Implement cycle to stabilize multi-step agent workflows.

Yes, you can reduce token waste by applying a diagnostic framework with falsification tests to evaluate whether specific context is actively needed or harmful before it enters the model window.

A Context Manifest template is an operational playbook structuring deliberate context sources for product managers and AI teams, needed when converting unfocused context stuffing into engineered memory architectures for decision-making.