memory

Index and search markdown and text files using SQLite FTS5.

3|1|Updated Mar 6, 2026

Search Tags:#context #indexing #recall #search #memory #sqlite

Ready to run

npx skills add https://github.com/mrap/hexagon-base --skill memory-mrap

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: memory
Source: https://github.com/mrap/hexagon-base/tree/main/dot-claude/skills/memory
Command: npx skills add https://github.com/mrap/hexagon-base --skill memory-mrap

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill provides a persistent, searchable memory for the AI agent, allowing it to recall past conversations, decisions, and project details, ensuring context is never lost.

Core Features & Use Cases

Full-Text Search: Indexes all markdown and text files for fast retrieval.
Contextual Recall: Enables the agent to answer questions about past interactions and information.
Use Case: When asked "What did we decide about the Q3 marketing budget last week?", the agent can query its memory to find the relevant decision and provide an accurate answer.

Quick Start

Use the memory skill to search for information about the last project meeting.

Frequently Asked Questions about memory

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

FAQPage Schema

How do I give an AI agent searchable memory for past conversations and decisions?▼

To give an AI agent searchable memory, you can index all markdown and text files within its directory using a local SQLite FTS5 database. This enables full-text search queries to recall past context, decisions, and project history.

How does full-text search indexing work for markdown and text files?▼

Full-text search indexing works by processing markdown and text files through Python scripts into a local SQLite FTS5 database. This allows the agent to quickly retrieve specific past interactions and project details through keyword queries.

Can I incrementally update the context index without reindexing all files?▼

Yes, you can incrementally update the context index without reindexing everything. The Python scripts support incremental updates for new information, while also offering a full reindexing option when a complete database rebuild is needed.

What is the best way to retrieve project history and people details from past interactions?▼

The best way to retrieve project history and people details is by querying the local SQLite FTS5 database. The agent searches its indexed markdown and text files to find and recall relevant past context and specific decisions.

Do I need Python to set up local SQLite full-text search for agent recall?▼

Yes, you need Python to set up this local SQLite full-text search for agent recall. The memory skill relies on Python scripts to handle both the indexing of markdown files and the execution of search queries.

Why does my agent lose context and forget past decisions from previous meetings?▼

An agent loses context and forgets past decisions because it lacks persistent memory. Indexing markdown files into a local SQLite database solves this by providing a searchable history of past interactions, decisions, and project details.