When should I use /explore instead of regular recall?

You should use /explore when your memory store is very large (over 1,000 items), when your question requires understanding the relationships between different parts of a system, or when standard search results aren't providing enough context.

What is RLM-style memory exploration?

RLM (Retrieval-Language Model) style exploration is an iterative process where the AI starts with a query, gathers initial hints, and then decides whether to peek at summaries or follow related nodes in a graph to find the most relevant information.

How does the Explore skill save on token costs?

Instead of loading 10+ full documents into the prompt, it uses lightweight 'peeks' (200-character summaries) and only 'expands' the specific documents it identifies as highly relevant through its exploration loop.

What are memory neighbors in this context?

Neighbors are related nodes in the memory graph connected by triplets (e.g., Node A -> uses -> Node B). This allows the skill to discover relevant information that might not share the same keywords but is functionally related.

Recursive Memory Explorer

Name: Recursive Memory Explorer
Author: genomewalker

bygenomewalker

데이터 과학 및 ML

Navigates complex memory graphs using RLM-style recursive search to find deep context without excessive token usage.

소개

The Recursive Memory Explorer skill implements a Retrieval-Language Model (RLM) style protocol for dynamically navigating large memory stores within Claude Code. Instead of relying on static top-k results that may miss associative context, this skill iteratively traverses a memory graph by peeking at summaries, following node relationships (triplets), and expanding only the most relevant content. This approach is specifically designed for large-scale repositories where understanding the connection between disparate components is critical, providing a transparent exploration trace while significantly reducing token consumption compared to standard full-context retrieval.

주요 기능

Iterative agent loop for autonomous context gathering
Token-efficient summary peeking before full expansion
Graph-based neighbor exploration via triplet connections
0 GitHub stars
Detailed exploration traces showing the search path and logic
RLM-style recursive memory navigation and discovery

사용 사례

Searching through memory stores exceeding 1,000 entries where top-k recall is insufficient
Tracing complex architectural relationships across interconnected system modules
Performing deep-dive research into specific implementation details within massive codebases

소개

주요 기능

Iterative agent loop for autonomous context gathering
Token-efficient summary peeking before full expansion
Graph-based neighbor exploration via triplet connections
0 GitHub stars
Detailed exploration traces showing the search path and logic
RLM-style recursive memory navigation and discovery

사용 사례

Searching through memory stores exceeding 1,000 entries where top-k recall is insufficient
Tracing complex architectural relationships across interconnected system modules
Performing deep-dive research into specific implementation details within massive codebases