Verified:GitHub

filesystem-context

Persist large tool outputs and multi-step plans to the filesystem.

1|Updated Jan 27, 2026

Search Tags:#workspace #context-engineering #agent-memory #tool-output #plan-persistence #scratch-pad #filesystem-context

Ready to run

npx skills add https://github.com/phonowell/mimikit --skill filesystem-context-phonowell

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: filesystem-context
Source: https://github.com/phonowell/mimikit/tree/main/.agents/skills/context-engineering-collection/skills/filesystem-context
Command: npx skills add https://github.com/phonowell/mimikit --skill filesystem-context-phonowell

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

Filesystem-based context engineering addresses token inflation and lost context by persisting large tool outputs, plans, and scratch space to the local filesystem, enabling durable agent memory across sessions.

Core Features & Use Cases

Scratch Pad Manager: offloads large tool outputs to disk and returns compact references for context.
Plan Persistence: saves and reloads multi-step plans to survive context window resets.
Sub-Agent Workspaces: enables isolated file-based communication between agents and a coordinator.
Dynamic Skill Loading: loads full skill content on demand to keep static prompts lean.
Terminal & Log Persistence: stores terminal outputs for targeted search and auditing.
Self-Modification Guard: provides safe persistence for learned preferences with guardrails.

Quick Start

Install and run a session that initializes the ScratchPadManager and AgentPlan to begin offloading large tool outputs and persisting multi-turn plans.

Frequently Asked Questions about filesystem-context

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

FAQPage Schema

How do I offload large tool outputs to disk to reduce token usage in agents?▼

You can offload large tool outputs to the filesystem using a ScratchPadManager, which persists data to disk and returns compact references to the agent context. This reduces token usage and preserves state for long-running tasks.

Why does my agent lose multi-step plans after a context window reset?▼

Agents lose multi-step plans after context window resets because state is held in volatile memory. Plan Persistence solves this by saving and reloading multi-step plans to the local filesystem, ensuring durable agent memory across sessions.

What is filesystem-based context engineering for durable agent memory?▼

Filesystem-based context engineering is a technique that persists large tool outputs, plans, and scratch space to the local filesystem. It enables durable agent memory across sessions and addresses token inflation and lost context.

Can I use isolated file-based communication between sub-agents and a coordinator?▼

Yes, you can use isolated file-based communication between sub-agents and a coordinator through Sub-Agent Workspaces. This pattern utilizes the filesystem to enable durable and structured inter-agent coordination.

Does dynamic skill loading keep static prompts lean for scalable agents?▼

Dynamic skill loading keeps static prompts lean by loading full skill content on demand. This approach reduces baseline token usage and supports scalable agent operations by fetching resources only when required.

What are the limitations of relying on filesystem context for agent state persistence?▼

Filesystem context persistence relies on local disk availability and requires managing file read/write permissions. While it preserves state across context window resets, it introduces disk I/O latency compared to purely in-memory agent state.