# Smart Memory

> Use when managing agent context, token limits, or designing memory architectures.

- Skill: `rushikeshsakharleofficial/smart-memory` (Agent Skill)
- Install (CLI): `npx skillmds@latest add rushikeshsakharleofficial/smart-memory`
- Raw SKILL.md: https://api.skillmd.com/api/skills/rushikeshsakharleofficial/smart-memory/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: rushikeshsakharleofficial (https://skillmd.com/u/rushikeshsakharleofficial)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/rushikeshsakharleofficial/smart-memory

---


# Smart Memory Management

Guide for agent memory. Goal = avoid context rot, save tokens, keep reasoning high.

## Memory Architecture (4 Tiers)

1. **Short-Term (Working):** Last 5-10 turns + active task state. In context window.
2. **Episodic:** Past interactions + tool results. Vector DB with recency decay.
3. **Semantic:** Knowledge base, facts, preferences (e.g., `MEMORY.md`, SQL).
4. **Procedural:** Core skills, persona, rules (system prompts).

## Context Optimization

- **Reduction:** Summarize history. Keep last N turns full, compress older turns recursively.
- **Offloading:** Move large tool outputs (HTML, big JSON) to disk. Keep structured pointer in context.
- **Selection:** RAG (Retrieval-Augmented Generation). Pull only relevant snippets.
- **Isolation:** Split tasks to sub-agents. Prevent context drowning.

## Advanced Tactics

- **Strategic Forgetting:** Prune noise (false starts, errors) after task done.
- **Structured Note-taking:** Use scratchpad (`NOTES.md`) for long-horizon state tracking.
- **Tool Condensation:** Tools return pre-processed data (row counts, keys), not raw payloads.
- **Prompt Caching:** Cache stable system prompts + large reference docs.

## Implementation Limits

- **Threshold Compaction:** Trigger summarize/offload when context hits 80-90%.
- **Recoverability:** Test if agent can find needle-in-haystack after compression.