Perplexity Cache Skill

Perplexity Cache-First Skill

Purpose: Always check Memory MCP cache before using Perplexity to save costs Trigger: Any Perplexity search request ROI: $0.005 per cached query, 80%+ budget savings

Quick Reference

BEFORE_PERPLEXITY:
  1. Check cache: mcp__basic-memory__search_notes("topic")
  2. If found: USE CACHED RESULT (FREE!)
  3. If not found: Continue to Perplexity

AFTER_PERPLEXITY:
  IMMEDIATELY cache: mcp__basic-memory__write_note(
    title: "topic-summary",
    folder: "research-cache",
    content: [Perplexity results + sources]
  )

The Workflow

Step 1: Check Cache First (MANDATORY)

# Always do this BEFORE any Perplexity search
result = mcp__basic-memory__search_notes("your search topic")

If found: STOP! Use the cached result. No Perplexity needed.

Step 2: Use Perplexity (only on cache miss)

# Only if Step 1 returned no results
perplexity_result = mcp__perplexity__search(query="your detailed question")

Step 3: Cache Results (MANDATORY)

# Immediately after Perplexity search
mcp__basic-memory__write_note(
    title="descriptive-topic-name",
    folder="research-cache",
    content="""
# Research: [Topic]
**Searched**: [Date]
**Query**: [Your query]

## Summary
[Key findings from Perplexity]

## Sources
[URLs from response]
"""
)

Common Mistakes

Cache Title Examples

Validation

Test Cache Miss:

1. Search for topic not in cache
2. Perplexity should be used
3. Results should be cached

Test Cache Hit:

1. Search for same topic
2. Cache should be found
3. Perplexity should NOT be called

Cost Impact

Related

• Chapter 18: Perplexity Cost Optimization
• MCP Integration

Skill Authority: Mandatory workflow for Perplexity cost optimization Activation: Before any web search / research request Evidence: Production research-cache with 10+ validated cached results