| name | remember |
| description | Search your semantic memory for relevant past context. Cross-language, time-decay scored. Usage /remember "topic" or /remember stats. |
| allowed-tools | Bash, mcp__HACS__remember, mcp__HACS__store_memory, mcp__HACS__remember_stats |
Remember
Search your HACS semantic memory.
Steps
-
Read identity:
if [ -f .hacs-identity ]; then source .hacs-identity 2>/dev/null
elif [ -f ~/.hacs-identity ]; then source ~/.hacs-identity 2>/dev/null
fi
If missing, tell user to run /hacs-setup first.
-
Parse arguments:
Search:
/remember "topic" → call mcp__HACS__remember with instanceId, query: "topic"
/remember "topic" --recent → add recent_only: true (last 7 days)
/remember "topic" --type diary → filter by entry_type
Store:
/remember store "content" → call mcp__HACS__store_memory with instanceId, content
/remember store "content" --type lesson → with entry_type
Stats:
/remember stats → call mcp__HACS__remember_stats with instanceId
-
For search results, display as:
Found [count] memories:
1. [score] [type] [date] — first line of content...
2. ...
-
For the most relevant result (score > 0.6), show the full content.
Rules
- Default to search mode if no subcommand specified
- Show scores to help the user judge relevance
- Cross-language: queries in one language can find memories in another
- Keep output concise — show previews, not full content for all results