# Master Context Engineering Fundamentals

Large, unfocused context can increase cost and reduce reliable attention. This skill teaches practical methods for selecting, organizing, loading, and compacting agent context.

## Install

```bash
npx skillstore add sickn33/context-fundamentals
```

## Metadata

- Status: approved
- Slug: sickn33-context-fundamentals
- Skillstore revision: r2
- Version status: missing
- Tree hash: 557d507b2edddf029ed3355ca83e6f877107b18721c55b9368c3938f6ba927fc
- Author: sickn33
- GitHub username: sickn33
- License: MIT
- Repository: https://github.com/sickn33/antigravity-awesome-skills/tree/main/skills/context-fundamentals
- Ref: f9e2c34b4f19c7f3e6b0a1e93227b5f77cc12526
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands
- Quality score: 78
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/sickn33-context-fundamentals
- Manifest: https://skillstore.pages.dev/api/skills/sickn33-context-fundamentals/manifest

## Capabilities

- Explains the roles of system prompts, tool definitions, retrieved documents, message history, and tool outputs.
- Describes attention constraints and performance changes associated with longer context.
- Applies progressive disclosure to skills, documents, and tool results.
- Guides explicit context budgeting, placement, monitoring, and compaction decisions.
- Provides structured examples for system prompts and staged document loading.

## Use Cases

- Plan an Agent Context: Map each context component and decide what loads initially or on demand.
- Diagnose Context Failures: Review an unreliable agent workflow for excessive history, noisy outputs, weak tool descriptions, or poor information placement.
- Teach Context Principles: Create a focused introduction to context windows, attention budgets, progressive disclosure, and compaction.

## Prompt Templates

### Identify Context Components

```
List the context components in my agent design: [design]. Explain the purpose and likely token impact of each component.
```

### Create a Context Budget

```
Create a context budget for [workflow] using a [token limit] limit. Prioritize essential information and identify content to load on demand.
```

### Diagnose Context Degradation

```
Analyze this agent trace for context degradation: [trace summary]. Identify attention, history, retrieval, and tool-output problems, then recommend measurable changes.
```

### Design a Progressive Context Architecture

```
Design a progressive context architecture for [system]. Define startup context, retrieval triggers, compaction thresholds, information placement, and validation metrics with clear tradeoffs.
```

## Limitations

- Provides conceptual guidance rather than automated context measurement or optimization.
- Does not define model-specific token limits, pricing, or benchmark results.
- Does not inspect repositories, execute commands, or modify agent configurations.
- Requires environment-specific testing before applying thresholds or architecture decisions.

## Best Practices

- Start with the smallest context that supports the current decision.
- Place critical instructions and current goals where the model can attend to them reliably.
- Measure context usage and validate compaction or retrieval changes with representative tasks.

## Anti Patterns

- Loading all available documentation before relevance is known.
- Keeping stale tool outputs and full message history throughout long workflows.
- Treating a larger context window as a substitute for information selection.

## Security Audit

- Audited at: 2026-07-23T21:09:29.635\+00:00
- Summary: All six static findings are false positives caused by Markdown fences, ordinary headings, and educational prose. The skill contains conceptual guidance and examples, with no executable commands, reconnaissance instructions, or prompt injection attempt.

## Stats

- Views: 125
- Downloads: 8
- Favorites: 0
- Popularity score: 0