# Principle Guard The Context Window

> Bound large outputs and broad research, delegate independent slices when available, and retain only evidence that can change the next decision.

- Skill: `njs14/principle-guard-the-context-window` (Agent Skill)
- Install (CLI): `npx skillmds@latest add njs14/principle-guard-the-context-window`
- Raw SKILL.md: https://api.skillmd.com/api/skills/njs14/principle-guard-the-context-window/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: njs14 (https://skillmd.com/u/njs14)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/njs14/principle-guard-the-context-window

---


# Protect useful context

Treat the request text that activated this skill as the work whose context use must be bounded.

Apply [guard the context window](../pkstack-principles/references/catalog.md#guard-the-context-window).
Define the next decision, read only evidence relevant to it, and route independent bulk research to
bounded native Kiro sub-agents when available. Keep compact findings rather than raw payloads.
Return the scope, summaries retained, and material evidence intentionally deferred.