research-logger_skill

Overview

This skill implements a search-and-auto-save research pipeline that logs every query and result to a local SQLite database while capturing full Langfuse traces. It uses Perplexity for search, stores entries with topic and project metadata, and preserves a complete audit trail of research sessions. The aim is to make research reproducible and never lose a session again.

How this skill works

When you run a search, the skill calls Perplexity to retrieve results and automatically persists the query, raw results, metadata (topic, project, tags), and timestamps to SQLite. Each operation is traced through Langfuse so you can inspect the request/response lifecycle, model behavior, and downstream actions. Simple CLI commands let you log new queries, search past entries, and view recent activity, enabling structured recall and analysis.

When to use it

• Conducting reproducible research that must be recalled later
• Building or updating a knowledge base from repeated searches
• Performing competitive analysis and tracking evidence over time
• Creating an audit trail for research decisions and citations
• Handing off research context to teammates or stakeholders

Best practices

• Always provide a clear topic and project name when logging searches to improve retrieval
• Use tags for concise categorization and faster filtering of results
• Secure the SQLite file and limit access to Langfuse keys and Perplexity credentials
• Run periodic exports or backups of the SQLite database to avoid single-point failure
• Review Langfuse traces for unexpected model behavior or debugging search issues

Example use cases

• Log market research queries and attach project metadata for later reporting
• Build a searchable knowledge base by periodically importing logged search results
• Compare vendors or technologies and keep timestamped evidence for decisions
• Audit past research sessions to reproduce findings and sources during reviews
• Share a project’s research history with teammates for fast onboarding