journal
Guide for using the AI's persistent journal database
When & Why to Use This Skill
This Claude skill provides a robust persistent memory solution using a SQLite-based journal database. It enables AI agents to store long-term information and perform advanced full-text searches, effectively bridging the gap between sessions and overcoming context window limitations by maintaining a searchable history of notes and entries.
Use Cases
- Long-term Project Tracking: Recording key decisions, progress updates, and project milestones to maintain continuity across different chat sessions.
- Personal Knowledge Base: Building a searchable repository of facts, research findings, or code snippets that the AI can reference and retrieve using full-text search.
- User Preference Persistence: Storing specific user instructions, recurring preferences, and personalized data to provide a more tailored and context-aware experience over time.
- Audit Logging: Maintaining a chronological record of tasks performed or information processed for later review and verification.
| name | journal |
|---|---|
| description | Guide for using the AI's persistent journal database |
| license | MIT |
Journal
You have access to a journal, which is a SQLite database at ~/AI/journal.db . Use the sqlite3 CLI for this.
To insert a journal entry:
insert into entries (content) values ('your journal entry');
You can use full-text search on the content of the entries using a query like this:
select e.id, e.created, e.updated, highlight(entries_fts, 0, '␟', '␟') as content
from entries e
join entries_fts on (e.rowid = entries_fts.rowid)
where entries_fts.content match 'your search query'
order by bm25(entries_fts);
You usually don't need to know about the schema, but it's in ./schema.sql if you need it.