Memory · Types

Procedural

The agent's rules of operation, instincts and "motor skills": model weights, code, and system prompt.

At a glance

ClassMemory layer

ScopePersistent (rules / skills)

ExamplesSystem prompts · skills · reflection

How it works

Reflection / meta-prompting: the agent evaluates its instructions against recent logs + user feedback and rewrites its own system prompt to improve future performance: an autonomous self-improvement loop.

Watch out

Agents rarely touch weights/code; self-rewriting prompts can drift without guardrails.

Basis

Internal research (Scalarity), Architectural Frameworks for AI Agent Memory Systems