Which databases does it interact with?

It mirrors data from the standard goose sessions database into a specialized reafference.db file located in the user's application support directory.

How does this skill help with cross-session awareness?

By mirroring session data into a dedicated reafference database, it allows the agent to 'find itself' in history and retrieve context from previous conversations, regardless of when they occurred.

What is Goose Introspection?

It is a specialized skill that allows Goose-based AI agents to access and query their own session history via a DuckDB database, enabling better continuity and self-discovery.

Can I use this to track token usage over time?

Yes, the skill includes views and tables specifically designed to track total tokens, message counts, and provider distribution across all tracked sessions.

Goose Session Introspection

Name: Goose Session Introspection
Author: plurigrid

byplurigrid

•

分析与监控

Enables cross-session awareness and agent self-discovery using a DuckDB-powered reafference database.

Goose Introspection provides a sophisticated framework for AI agents to maintain historical context and self-awareness across multiple sessions. By implementing a reafference-based discovery system, it allows Claude and Goose instances to query their own past interactions, track evolution across different providers or models, and perform deep analysis of session metadata. This is particularly useful for long-term projects where maintaining continuity and understanding previous decision-making patterns is critical for agent performance.

主要功能

01Automated 'self-discovery' allowing agents to locate their current session in the timeline

02DuckDB-backed persistence for cross-session history and message querying

03Searchable session views for content analysis and interaction entropy tracking

04Multi-provider tracking for sessions across Anthropic, OpenAI, and Google models

05Comprehensive session analytics including token usage, message counts, and provider distribution

062 GitHub stars

使用场景

01Auditing agent behavior and token efficiency across hundreds of historical sessions

02Maintaining continuity in long-term development projects when switching between different LLMs

03Implementing 'reafference' testing to help agents predict and validate their own actions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add plurigrid/asi goose-introspection

For use in Claude.ai and ChatGPT

Download Skill