mcpField
S
supermemory-search

supermemory-search

github.com/supermemoryaiAgent workflows2.8k93 installsUpdated 1d ago
Repository

About supermemory-search

A Claude skill for searching Supermemory for past coding sessions, decisions, and saved information. It supports searches across personal memories, project memories, or both, and presents results with timestamps and relevance scores.

Key features

  1. 01Searches past sessions and decisions
  2. 02Scopes searches to user or project
  3. 03Shows timestamps and relevance scores

Use cases

  1. 01Finding recent coding activity
  2. 02Looking up a project implementation decision
  3. 03Checking saved coding preferences

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/supermemoryai/claude-supermemory --skill supermemory-search

After installing, type/supermemory-searchin Claude Code to confirm it loaded.

Details

From the project README

Supermemory Search

Search Supermemory for past coding sessions, decisions, and saved information.

How to Search

Run the search script with the user's query and optional scope flag:

node "${CLAUDE_PLUGIN_ROOT}/scripts/search-memory.cjs" [--user|--repo|--both] "USER_QUERY_HERE"
Scope Flags
  • --both (default): Search both personal session and project memories across team members in parallel
  • --user: Search personal/user memories across sessions
  • --repo: Search project/repo memories across team members

Examples

  • User asks "what did I work on yesterday":

    node "${CLAUDE_PLUGIN_ROOT}/scripts/search-memory.cjs" "work yesterday recent activity"
    
  • User asks "how did we implement auth" (project-specific):

    node "${CLAUDE_PLUGIN_ROOT}/scripts/search-memory.cjs" --repo "authentication implementation"
    
  • User asks "what are my coding preferences":

    node "${CLAUDE_PLUGIN_ROOT}/scripts/search-memory.cjs" --user "coding preferences style"
    

Present Results

The script outputs formatted memory results with timestamps and relevance scores. Present them clearly to the user and offer to search again with different terms if needed.

Comments

Markdown supported