Most agents hit context limits faster than you'd expect, and this skill treats the filesystem as overflow storage instead of cramming everything into the prompt. Write tool outputs to scratch files, persist plans across long tasks, let sub-agents communicate through shared directories, and load skill instructions dynamically instead of bloating your system prompt. The core insight is simple: static context is expensive real estate, so keep pointers in the prompt and pull full content on demand with grep and targeted reads. Works well for terminal output persistence and cross-session state. The self-modifying agent pattern (writing learned preferences back to instruction files) is interesting but needs guardrails. If you're building anything that runs longer than a few turns or coordinates multiple agents, you'll need this approach.

skills.sh installs
378
GitHub stars
18.3K
last updated
May 26, 2026

Install

$ npx -y skills add muratcankoylan/agent-skills-for-context-engineering --skill filesystem-context --agent claude-code

Installs into .claude/skills of the current project.

Files

Filesystem-Based Context Engineering

Use the filesystem as the primary overflow layer for agent context because context windows are limited while tasks often require more information than fits in a single window. Files let agents store, retrieve, and update an effectively unlimited amount of context through a single interface.

Prefer dynamic context discovery -- pulling relevant context on demand -- over static inclusion, because static context consumes tokens regardless of relevance and crowds out space for task-specific information.

When to Activate

Activate this skill when:

  • Tool outputs are bloating the context window
  • Agents need to persist state across long trajectories
  • Sub-agents must share information without direct message passing
  • Tasks require more context than fits in the window
  • Building agents that learn and update their own instructions
  • Implementing scratch pads for intermediate results
  • Terminal outputs or logs need to be accessible to agents

Sponsors

Details

Repository
github.com/muratcankoylan/agent-skills-for-context-engineering
Category
AI & Agent Building
License
MIT
First seen
Jun 11, 2026

More from muratcankoylan/agent-skills-for-context-engineering

All 17
1davila7 avatarsentence-transformersdavila7/claude-code-templatesFramework for state-of-the-art sentence, text, and image embeddings. Provides 5000+ pre-trained models for semantic similarity, clustering, and retrieval. Suppo376skills.sh installs
2sanity-io avatardial-your-contextsanity-io/agent-contextdial your context376skills.sh installs
3syncfusion avatarsyncfusion-blazor-smart-pastesyncfusion/blazor-ui-components-skillsSet up and use the Syncfusion Blazor Smart Paste Button — AI-powered clipboard-to-form filling with OpenAI, Azure OpenAI, Ollama, or custom AI backends. Covers 376skills.sh installs
40xindiebruh avataropenclaw-mission-control0xindiebruh/openclaw-mission-control-skillCoordinate AI agent teams via a Kanban task board with local JSON storage. Enables multi-agent workflows with a Team Lead assigning work and Worker Agents execu371skills.sh installs
5syncfusion avatarsyncfusion-blazor-smart-textareasyncfusion/blazor-ui-components-skillsSet up and configure the Syncfusion Blazor Smart TextArea for AI-powered inline or popup text autocompletion. Covers OpenAI, Azure OpenAI, Ollama, and custom IC371skills.sh installs
6liangdabiao avatarstock-question-refinerliangdabiao/claude-code-stock-deep-research-agent股票投资调研问题细化技能。将用户提供的股票名称/代码细化为结构化的8阶段投资尽调指令。通过提问澄清投资风格(价值/成长/困境反转)、持有周期(短/中/长线)、风险偏好、研究重点,生成符合专业投资研究标准的结构化调研任务。当用户提到股票分析、投资研究、股票尽调时使用此技能。369skills.sh installs