What is the primary purpose of the Active Interleave skill?

It retrieves and integrates context from your most recent Claude and Amp threads into your current session to improve task continuity and accuracy.

How does the random walk feature work?

It navigates a graph of your interaction history, known as awareness edges, to find and sample related fragments of information that are mathematically relevant to your current task.

Can I limit the context search to a specific timeframe?

Yes, the skill supports an --hours flag, allowing you to filter the context window to a specific duration, such as the last 6 or 24 hours of work.

Do I need a specific database to use this skill?

Yes, the skill reads from a local DuckDB instance located at ~/worldnet/cognition.duckdb which stores session metadata, messages, and temporal indexes.

What is GF(3) balanced exploration?

It is a mathematical framework used by the skill to ensure information is validated (Minus), explored (Ergodic), and generated (Plus) in a coherent, balanced manner satisfying conservation laws.

Active Interleave

Name: Active Interleave
Author: plurigrid

byplurigrid

•

생산성 및 워크플로

Synchronizes and surfaces relevant context from recent Claude and Amp sessions using topological random walks to enhance current task performance.

The Active Interleave skill implements advanced active inference patterns to bridge the gap between separate working sessions. By performing random walks over an awareness graph stored in a local DuckDB cognition database, it retrieves and integrates fragmented context from your most recent activity. This allows Claude to maintain continuity across disjointed tasks, minimize uncertainty by foraging for epistemic fragments, and leverage insights from recent threads without manual intervention, all while adhering to GF(3) balanced exploration principles for compositional coherence.

주요 기능

01DuckDB Cognition Integration: Directly queries session metadata and awareness edges from the worldnet cognition store.

02Time-Based Filtering: Enables specific windowing of context retrieval, such as the last 6 or 24 hours of activity.

03GF(3) Coherence: Utilizes balanced triadic exploration (MINUS/ERGODIC/PLUS) for stable and mathematical information synthesis.

04Query-Focused Extraction: Supports targeted context interleaving based on specific keywords or domain topics.

05Contextual Random Walk: Samples recently active threads to discover relevant task fragments across the awareness graph.

068 GitHub stars

사용 사례

01Reducing cognitive load during deep work by automatically surfacing relevant epistemic fragments from recent history.

02Cross-referencing logic and API decisions made in parallel development threads without switching contexts.

03Resuming a complex development project by pulling in architectural context from the previous day's troubleshooting sessions.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add plurigrid/asi active-interleave

For use in Claude.ai and ChatGPT

Download Skill