Can I limit results to only successful past tasks?

Yes, the skill supports verdict filtering, allowing you to retrieve only successful episodes to ensure the AI follows proven paths.

How is the retrieved context formatted?

It returns a structured RetrievalContext object containing episode summaries, key steps, reflections, artifacts, and relevance scores for easy consumption by the AI model.

How does the Context Retrieval skill find relevant information?

It primarily uses semantic search with vector embeddings to find similar intents. If embeddings are unavailable, it falls back to high-performance SQL keyword and tag indexing.

What storage systems are required for this skill?

The skill is designed to work with a hybrid setup using Turso (SQL) for scalable long-term storage and redb (KV) for fast local caching.

Does this skill help with code generation patterns?

Absolutely. It can retrieve 'ToolSequence' patterns which provide recommended sequences of actions based on what has worked best in the past for similar domains.

Context Retrieval for AI Memory

Name: Context Retrieval for AI Memory
Author: d-o-hub

byd-o-hub

•

データサイエンスとML

Retrieves relevant past episodes and patterns from episodic memory to inform AI decision-making and task execution.

The context-retrieval skill empowers AI agents by providing access to a modular Rust-based self-learning memory system. It allows Claude to draw upon historical data, including past successes, failures, and implementation patterns, to make more informed decisions during complex workflows. By leveraging semantic search via embeddings or deterministic keyword indexing through Turso and redb, the skill ensures that relevant context—such as heuristics, error resolutions, and proven tool sequences—is always available to optimize current task performance and avoid repeating past mistakes.

主な機能

01Fallback keyword and tag-based indexing for fast, deterministic results

02Multi-dimensional filtering by domain, task type, recency, and success rate

03Hybrid storage retrieval utilizing redb for caching and Turso for persistent SQL data

04Structured data responses including episode summaries, reflections, and heuristics

053 GitHub stars

06Semantic search using vector similarity for intent-based context recall

ユースケース

01Identifying proven tool sequences for specific architectural patterns in storage systems

02Reviewing project reflections to ensure current implementations align with learned best practices

03Solving recurring Rust compiler errors by retrieving successful past resolutions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add d-o-hub/rust-self-learning-memory context-retrieval

For use in Claude.ai and ChatGPT

Download Skill