Episodic vs Semantic Memory

Episodic memory stores specific past experiences and interactions (what happened during a particular coding session, how a specific bug was resolved), while semantic memory stores general knowledge and facts (project conventions, API patterns, architectural decisions) — and the distinction determines how agents learn from experience versus encode persistent knowledge. Episodic memories are time-stamped, contextual, and decay naturally — they're most useful for recalling "that one time the agent tried the wrong approach and had to backtrack," enabling the agent to avoid repeating mistakes. Semantic memories are abstracted, generalized, and persistent — they represent the distilled lessons from many episodes, like "always run tests before committing" or "this project uses camelCase." The most effective agent memory systems convert episodic memories into semantic ones over time: individual debugging experiences become general error-handling heuristics, and specific code review feedback becomes generalized style rules. This concept connects to memory types for the broader taxonomy, long-term memory for persistence mechanisms, memory management for governing the lifecycle of both memory types, and agent config files as a lightweight form of semantic memory that persists across sessions.