graphiti-memory-retrieval

# Graphiti Memory Retrieval

## Description

Techniques for searching and retrieving information from Graphiti Memory. Provides proven search patterns for finding similar past work, error solutions, node summaries, and learning from historical data.

## When to Use

Automatically apply this skill when:
- Starting tasks similar to previous work
- Encountering errors that may have been seen before
- Looking for past patterns and solutions
- Need to understand entity relationships
- Building on proven approaches
- Analyzing trends across sessions

## Core Search Types

**Facts Search**: Find specific relationships between entities
- "Agent X solved problem Y using approach Z"
- "Configuration A causes error B"
- "Pattern C appears in successful projects"

**Node Search**: Get comprehensive entity summaries
- All relationships for an entity
- How entities connect across episodes
- Holistic view of patterns

## Search Patterns

### Pattern 1: Search for Similar Past Work

Before starting, search for similar tasks:

```python
# Search for similar operations
mcp__graphiti-memory__search_memory_facts(
    query="FastAPI async database connection setup with PostgreSQL",
    group_ids=["python_development", "agent_executions"],
    max_facts=5
)

# Use results to:
# - Understand approach that worked before
# - Avoid past mistakes
# - Estimate time based on previous similar work
```

**When to use**: Before starting new work
**Value**: Learn from past successes, avoid repeating mistakes

### Pattern 2: Search for Error Patterns

When encountering errors, search knowledge base:

```python
# Search for similar errors
mcp__graphiti-memory__search_memory_facts(
    query="PostgreSQL connection pool exhausted timeout",
    group_ids=["error_resolutions"],
    max_facts=3
)

# Use results to:
# - Apply known solutions
# - Avoid trying failed approaches
# - Understand root causes faster
```

**When to use**: When encountering errors
**Value**: Faster resolution using known solutions