read-memories

Search JSONL agent session logs with DuckDB queries to recover past decisions.

1|Updated Apr 3, 2026
One-click install
npx skills add https://github.com/nq-rdl/agent-extensions --skill read-memories
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: read-memories
Source: https://github.com/nq-rdl/agent-extensions/tree/main/skills/duckdb/read-memories
Command: npx skills add https://github.com/nq-rdl/agent-extensions --skill read-memories

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Search past agent session logs stored as JSONL to recover decisions, patterns, or unresolved work. Use when the user references past conversations, asks what was done before, or wants context recovered from local transcript files.

Core Features & Use Cases

  • Silent retrieval of prior decisions and patterns from local logs without narrating the lookup process.
  • Context enrichment: absorbs results and continues with enriched context for the current task.
  • Use Case: When a user asks what happened earlier in a project, this skill surfaces relevant transcripts and decisions.

Quick Start

Provide a keyword to search; the skill will locate the most relevant transcripts and return summarized results.

Frequently Asked Questions about read-memories

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

FAQPage Schema
How do I search past agent session logs to recover previous decisions?

Recover context from past conversations by querying local JSONL session logs with DuckDB. The skill silently searches transcript files for matching keywords and returns relevant prior decisions or unresolved work to enrich your current task.

What is context recovery from local session transcripts?

Context recovery from local transcripts is querying JSONL session logs to extract past decisions and unresolved work. It searches historical records using DuckDB and absorbs matching results to seamlessly enrich the active agent session.

How do I query JSONL log files with DuckDB to find specific conversation history?

Query JSONL log files with DuckDB by providing a keyword and a valid log root path. The skill applies path-resolution logic and properly escaped SQL to search transcripts, returning matching conversation history and summarized results.

Do I need a specific log directory structure to search session logs?

Yes, searching session logs requires a valid log root directory containing JSONL files. The skill applies path-resolution logic to navigate this structure and locate relevant transcripts across your local projects for accurate recovery.

Can I find unresolved work from previous agent sessions across different projects?

Yes, you can find unresolved work across projects by querying local JSONL session logs with DuckDB. Provide a keyword to search transcript files, and the skill recovers past patterns or incomplete tasks to enrich current context.