From mneme
Store a memory in mneme — use when you learn a preference, knowledge, or decision worth remembering
How this command is triggered — by the user, by Claude, or both
Slash command
/mneme:rememberThis command is limited to the following tools:
The summary Claude sees in its command listing — used to decide when to auto-load this command
# mneme remember
Store a memory in mneme's searchable memory system. Use this whenever you learn something worth remembering — preferences, knowledge, decisions — from the conversation.
## When to use
- The user states a preference ("always use X", "never do Y", "use bun not npm")
- You learn project knowledge ("this project uses stoolap for storage", "CI runs on Blacksmith")
- A decision is made ("we chose Catppuccin Mocha for dark theme", "memory feedback is idempotent")
- The user explicitly asks you to remember something
## Instructions
Determine the appropriate type and content, t...Store a memory in mneme's searchable memory system. Use this whenever you learn something worth remembering — preferences, knowledge, decisions — from the conversation.
Determine the appropriate type and content, then run:
mneme memory add "<content>" -t <preference|knowledge|decision>
Memories are automatically scoped to the current project based on your working directory. The mneme server resolves your CWD against registered paths to determine the scope. No --project or --scope flags are needed.
To store a global memory (applies across all projects), add --scope-global:
mneme memory add "<content>" -t <preference|knowledge|decision> --scope-global
| Signal | Type |
|---|---|
| "always", "never", "prefer", "use X not Y", "I like", "don't" | preference |
| Facts, how things work, architecture, tooling, conventions | knowledge |
| "we decided", "the approach is", trade-off resolutions | decision |
| Scenario | What happens |
|---|---|
| Running from a project directory | Memory auto-scoped to the closest registered path |
--scope-global flag used | Memory stored as global (no path association) |
| CWD has no registered ancestor path | Memory stored as global |
User says: "this project uses bun exclusively, use bun/bunx instead of npm/npx"
mneme memory add "This project uses bun exclusively as its JavaScript runtime. Always use bun/bunx instead of npm/npx." -t preference
User says: "always use rg instead of grep" (applies everywhere):
mneme memory add "Prefer using rg (ripgrep) instead of grep for faster searches" -t preference --scope-global
This command stores in mneme's memory system — searchable, scoreable, and visible in the web UI. This is in addition to Claude Code's built-in auto-memory (file-based). Both should be used: auto-memory for Claude's own context, mneme for the shared knowledge base.
npx claudepluginhub sentiolabs/claude-marketplace --plugin mneme/rememberLogs a finding or successful exploit pattern to persistent hunt memory, auto-populating from session context and /validate output.
/rememberAdds user-specified content as a manual entry to .claude/memories/project_memory.json and confirms with a success message.
/rememberManually capture a memory now (independent of Stop-hook auto-capture). Pass -p <path> to skip classification.
/rememberStores decisions, patterns, outcomes, and context in a knowledge graph for future reference. Supports flags for success/failure tracking, categories, agent scoping, and global best practices.
/rememberSaves input as a classified experience (solution, gotcha, pattern, technique, decision, preference) to a JSON file in _memory/experiences/, extracting summary, problem, solution, and tags. Confirms with summary.