context-saver

Index full skill outputs with FTS5 while returning compact summaries.

12|5|Updated Mar 14, 2026
One-click install
npx skills add https://github.com/tlancas25/context-cooler --skill context-saver-tlancas25
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: context-saver
Source: https://github.com/tlancas25/context-cooler/tree/main
Command: npx skills add https://github.com/tlancas25/context-cooler --skill context-saver-tlancas25

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @modelcontextprotocol/sdk, better-sqlite3, turndown, zod, and includes scripts (resource) components.

What problem does it solve?

Context Saver dramatically reduces the data volume entering the AI context by executing data-heavy tasks in a sandbox and returning compact summaries, while indexing the full outputs for on-demand retrieval.

Core Features & Use Cases

  • Safe, deterministic sandboxed execution of skill commands across languages.
  • Intent-driven filtering and field selection to produce concise summaries (100-500 bytes) for context windows.
  • Session continuity via SQLite-based events and 2KB snapshots to preserve state across conversations.
  • Batch execution of multiple skills with automatic indexing for search and traceability.
  • Automated installer that wires Context Saver into AGENTS.md, TOOLS.md, and cron jobs.

Quick Start

Run a data-heavy skill through Context Saver to obtain a compact summary while the full data is indexed for later search.

Frequently Asked Questions about context-saver

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

FAQPage Schema
How do I reduce token usage when running data-heavy operations in an AI context window?

To reduce token usage for data-heavy operations, you can execute skills in a sandboxed environment that returns compact 100-500 byte summaries to the context window while indexing full outputs for on-demand search.

How does FTS5 indexing work with sandboxed skill execution?

FTS5 indexing captures full output data from sandboxed skill executions into SQLite, enabling on-demand search and retrieval of detailed results without loading the entire dataset into the context window.

Can I maintain session continuity across multiple AI conversations using SQLite?

Yes, session continuity is maintained by storing events and 2KB state snapshots in SQLite, preserving context across conversations while keeping active token consumption minimal.

What is the best way to batch process multiple skills while minimizing context window load?

Batch processing multiple skills is handled by executing them in a sandboxed environment with automated indexing, returning concise summaries to the context window while preserving full data for traceability.

Does Context Saver automatically integrate with AGENTS.md and cron jobs?

Yes, an automated installer wires the skill into AGENTS.md, TOOLS.md, and cron jobs, facilitating scheduled sandboxed execution and intent-driven filtering without manual configuration.