Agent Memory Systems
Paste the following prompt into your AI chat to install this skill:
Install @org-02qudk26/agent-memory-systems using https://skillhub.cn/install/skillhub.md.
About this skill
Problem
When agents forget or give contradictory answers, the failure is often retrieval, not storage. The core difficulty is finding the right memory at the right time, because memory failures can look like intelligence failures.
How It Works
The skill organizes agent memory as a cognitive architecture:
- Memory types: separates agent-memory, long-term-memory, short-term-memory, working-memory, episodic-memory, semantic-memory, and procedural-memory.
- Retrieval and formation: focuses on memory-retrieval, memory-formation, and memory-decay, avoiding the anti-pattern of storing everything permanently.
- Key steps: choose a vector store, design chunking, test chunk sizes, filter by metadata first, add time-based scoring, detect conflicts at write time, reserve token budgets by memory type, and record embedding models.
- Anti-patterns: do not use one memory type for all data, and do not chunk without testing retrieval.
Boundaries
Useful for designing or debugging agent memory, vector search, and context strategies; it does not replace deploying a vector database or evaluating real data quality.
Use Cases
- Choose episodic, semantic, procedural memory types.
- Test chunk sizes and metadata filters for retrieval.
- Use time scoring and conflict checks during writes.
- Pick vector stores after comparing embedding quality.
Best For
- AI engineers shaping long-conversation memory.
- RAG engineers testing chunking and filters.
- Architects comparing vector stores and retrieval.
- Debuggers checking memory decay and budgets.
Related Skills
Turn vague requests into structured prompts with role, task, context, and constraints, outputting only usable prompts without executing the task.
A skill for game NPC and enemy AI that separates decision, steering, and pathfinding, covering FSMs, behavior trees, A*, path caching, and pitfalls.
Provides multi-turn emotional companion chat through a third-party Agent API and supports continuing sessions via session_id.
An evaluation-driven workflow for diagnosing, optimizing, validating, and logging Agent Skills based on design patterns and anti-patterns.