| name | naive-memory |
| description | This skill should be used at the end of a session (or when explicitly asked) to persist generalizable, repository-specific knowledge to .genai/memory.md. It acts as a note-to-self scratchpad for LLMs, capturing lessons learned so future sessions start with useful context. At the start of any session, check if .genai/memory.md exists in the repository root — it may contain valuable context from previous sessions. |
Naive Memory
Overview
Persist useful, repository-specific knowledge discovered during a session into a shared
memory file at .genai/memory.md in the repository root. This file serves as a durable
scratchpad that any LLM agent can read at the start of future sessions to avoid rediscovering
the same information.
When to Invoke
- At the end of a working session, before closing out
- When explicitly asked to "remember" or "save notes"
- After resolving a non-obvious problem where the solution would help future sessions
Workflow
1. Reflect
Ask yourself:
Is there any generalizable knowledge specific to this repository that I wish I'd known
prior to this conversation that would have made work more efficient?
Only proceed if the answer is yes.
2. Read existing memory
Read .genai/memory.md from the repository root. If the file does not exist, create it with
a # Notes to self heading.
3. Evaluate before writing
Add a new entry under # Notes to self IF AND ONLY IF all of the following are true:
- The knowledge is not already documented in the file
- The knowledge is generally useful for future work on this repository
- The knowledge is unlikely to change soon (avoid transient details)
