How does the tiered retrieval system work?

The system detects available search capabilities and prioritizes the most effective method, such as Topic or Hybrid search, falling back to Keyword or manual TOC navigation if indices are missing.

Does this skill access current session memory?

No, this skill is specifically designed to query historical data and past sessions. Information from the current active session is already available in Claude's immediate context.

What is a Grip ID in the search results?

A Grip ID is a unique 13-digit timestamped identifier that allows the agent to precisely locate and expand a specific segment of conversation for better context.

Can I search across different projects?

No, memory stores are managed on a per-project basis to ensure that context remains relevant and isolated to the specific codebase you are currently working on.

Agent Memory Query

Name: Agent Memory Query
Author: SpillwaveSolutions

bySpillwaveSolutions

•

生産性とワークフロー

Retrieves and analyzes historical conversation data using intelligent tiered search and intent-based routing for long-term project context.

The Memory Query skill provides Claude with a persistent memory layer, allowing it to recall specific discussions, decisions, and code snippets from past sessions. By utilizing a sophisticated tiered retrieval system—incorporating semantic vector search, BM25 keyword matching, and hierarchical TOC navigation—it ensures high-precision retrieval even when specific indices are unavailable. This skill is essential for maintaining project continuity, enabling Claude to answer questions like 'what did we decide last week' or 'find the authentication logic we discussed' by automatically classifying query intent and applying time-boxed filters.

主な機能

014 GitHub stars

02Intelligent intent classification to optimize search strategy for browsing vs. locating

03Context expansion via Grip IDs to retrieve surrounding lines from historical excerpts

04Full explainability metadata for every query, including confidence scores and methods used

05Time-based hierarchical navigation through years, months, weeks, and days

06Tiered retrieval with automatic fallback chains (Semantic, Hybrid, Keyword, and Agentic)

ユースケース

01Recalling specific architectural decisions or debugging steps from previous development sessions

02Browsing historical project topics to understand the evolution of a codebase or feature set

03Locating exact snippets or configuration values mentioned in past conversations without manual scrolling

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add spillwavesolutions/agent-memory memory-query

For use in Claude.ai and ChatGPT

Download Skill