mem_skill

Overview

This skill lets you run a local-first memory search against a cached index to surface relevant archived skill data. It is designed for quick, private lookups and integrates with /mem queries in Telegram. The goal is fast retrieval of top hits with their file paths and headers for easy reference.

How this skill works

The skill maintains a local index of archived skill versions and performs searches against that index rather than querying external services. You can update the index when needed and run search queries that return the top matches with paths and headers; brief summaries are provided when useful. Searches run locally for low latency and privacy.

When to use it

• You need to find archived versions or metadata of skills quickly from a local cache.
• You want private, offline-capable searches for previously crawled skill data.
• Responding to /mem queries in Telegram where network calls should be minimized.
• Exploring historical changes or backups of skill files and headers.

Best practices

• Regularly update the local index before heavy lookups to include recent archives.
• Use concise, specific queries to get more precise top hits.
• Limit --top to a small number (3–10) when you need the most relevant results fast.
• Inspect returned file paths and headers directly for context before summarizing.
• Keep the index storage backed up if it holds important archival data.

Example use cases

• Search for all archived versions of a skill to compare changes across releases.
• Answer a Telegram /mem query by returning top matching files and a short summary.
• Locate a specific header or file path inside the archive when troubleshooting.
• Quickly retrieve examples or snippets stored in archived skill files for reference.