filesystem-context

Offloads tool outputs to timestamped files and persists agent plans as JSON for reuse across sessions.

Updated May 24, 2026
One-click install
npx skills add https://github.com/FVossebeld/agent-skills-for-context-engineering --skill filesystem-context-fvossebeld
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: filesystem-context
Source: https://github.com/FVossebeld/agent-skills-for-context-engineering/tree/main/skills/filesystem-context
Command: npx skills add https://github.com/FVossebeld/agent-skills-for-context-engineering --skill filesystem-context-fvossebeld

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

The Filesystem Context skill enables durable, file-backed storage for agent context, allowing scratch spaces, tool-output offloading, and cross-session handoffs to persist information beyond a single prompt.

Core Features & Use Cases

  • ScratchPadManager offloads large tool outputs to timestamped files and returns compact references to maintain a lean active context.
  • AgentPlan and PlanStep provide persistent, re-readable multi-step task tracking across context refreshes.
  • ToolOutputHandler automatically decides inline vs offload for varied tool outputs, integrating with lightweight skill loaders and sub-agent workspaces.
  • Optional sub-agent workspaces support coordinated multi-agent collaboration and state sharing through a common filesystem.

Quick Start

Use filesystem-context to offload a large tool output to scratch and then load the saved plan to continue work.

Frequently Asked Questions about filesystem-context

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I persist agent plans and context across context window refreshes?

You can persist agent plans across context refreshes by serializing multi-step tasks to JSON files on disk, allowing the agent to rehydrate plans and resume work seamlessly after losing active context.

What is the best way to offload large tool outputs from agent context?

Offload large tool outputs to timestamped filesystem scratch files, returning compact references to the agent to maintain a lean active context and prevent token limit exhaustion.

Does filesystem context storage work for cross-agent handoffs and multi-agent collaboration?

Filesystem context storage supports cross-agent handoffs by using optional sub-agent workspaces that allow coordinated multi-agent collaboration and state sharing through a common filesystem.

How do I automatically decide between inline tool output and filesystem offload?

Use a ToolOutputHandler to evaluate varied tool outputs and automatically decide whether to keep results inline or offload them to disk, integrating with lightweight skill loaders.

When should I not use filesystem-backed scratch space for agent memory?

Avoid filesystem-backed scratch space if your agent workflow operates within a single context window without multi-turn persistence needs, or if disk I/O latency negatively impacts performance.