← ClaudeAtlas

context-window-engineerlisted

Optimize 2M+ context windows, prompt cache pinning & budget slicing / Rekayasa context window 2M+, cache pinning, dan budget token.
roedyrustam/vibes-plug · ★ 126 · AI & Automation · score 90
Install: claude install-skill roedyrustam/vibes-plug
# context-window-engineer — vibes-plug Skill [English](#english) | [Bahasa Indonesia](#bahasa-indonesia) --- <a name="english"></a> ## English ### Orchestration & Integration - **Upstream & Orchestration**: `zero-to-prod-orchestrator` (Phase 4 AI integration), `brainstorming`, `session-memory-manager` (context zone snapshots & handoff). - **Prefix Locking & Model Cascading**: `kv-cache-prefix-optimizer` (Zone 1-2 cache boundary alignment), `adaptive-model-cascade` (overflow routing across model sizes), `llm-finops-router` (per-zone token cost accounting). - **Frontier LLMs & Model Boosters**: `frontier-ai-models-expert` (2M+ token recall curves), `gemini-agent-booster` (Google 2M+ native context caching), `vercel-ai-sdk-expert` (context streaming & tools). - **Research & Deep Ingestion**: `deep-research-analyst` (retrieval-augmented injection), `vector-db-rag-expert` (retrieval vs context stuffing decisions). ### Description Production guide for engineering, partitioning, and optimizing ultra-large context windows (128k → 2M+ tokens) across frontier models. Prevents the most common failure mode in 2026 AI applications: dumping raw data into context without structure, causing degraded recall, wasted tokens, and hallucination amplification in the "lost in the middle" zone. ### Trigger Conditions Activate this skill when: - Building AI agents or applications that consume >50k tokens of context per request. - Experiencing degraded model recall, "lost in the middle" failures