AgentSkillsCN

Memory Search

内存搜索

SKILL.md

Memory Search

Quick query of memories for specific topics.

What This Skill Does

  1. •Takes a search query as argument
  2. •Calls mcp__memory__search_nodes with the query
  3. •Displays matching entities and observations
  4. •Shows relevance/match quality
  5. •Offers to drill deeper or update

Usage

code
/memory-search trailing stops
/memory-search "PostgreSQL decision"
/memory-search committee quorum
/memory-search position management

Output Format

code
=== MEMORY SEARCH: "trailing stops" ===

MATCHES (2 found):

1. Position Management System [Trading Infrastructure]
   Relevance: ★★★★☆ (direct match)

   Matching observations:
   → "Trailing stops with dynamic percentages based on volatility"
   → "Scale-out functionality for profitable positions"

   Related entities:
   - Committee Tick Trading (uses positions)
   - Exchange Integration (executes stops)

2. 16 Agent Personality Traits [Feature]
   Relevance: ★★☆☆☆ (indirect)

   Matching observations:
   → "stop_loss_tightness: 0-1 scale for how tight stops are"

NO EXACT MATCH? Try:
- "stop loss" (related concept)
- "risk management" (broader category)
- "position" (parent concept)

Actions:
[D]rill into entity | [U]pdate observation | [A]dd new | [Q]uit

Search Tips

Query TypeExampleBest For
Single wordcommitteeBroad search
Phrase"voting weights"Exact match
Multiple wordsposition riskAND search
Questionhow does committee voteConceptual

Process

  1. •Parse query (handle quotes, multiple words)
  2. •Call mcp__memory__search_nodes(query)
  3. •Rank results by relevance:
    • •Entity name match: 100 points
    • •Entity type match: 50 points
    • •Observation match: 30 points per match
  4. •Display with relevance stars
  5. •Show related entities via relations
  6. •Offer actions (drill, update, add)

When to Use

  • •Quick lookup of specific topics
  • •Before asking a question (check if known)
  • •Finding related concepts
  • •Verifying memory contains expected info