graphiti-learning-workflows

# Graphiti Learning Workflows

## Description

Workflows for learning from historical data and building institutional knowledge over time. Integrates episode storage and memory retrieval into cohesive learning patterns that improve agent performance across sessions.

## When to Use

Automatically apply this skill when:
- Starting tasks similar to previous work
- Building knowledge over time
- Analyzing trends in development work
- Creating audit trails for compliance
- Improving from past successes and failures
- Recognizing patterns across sessions

## Learning Workflow Stages

### Stage 1: Before Starting Work

**Ask**: "Have I done something like this before?"

**Workflow**:
```
1. Search for similar tasks in relevant group_ids
2. Review past approaches and outcomes
3. Note lessons learned from previous work
4. Apply proven patterns
5. Use proven successful approaches
```

**Example**:
```python
# User asks: "Build REST API with JWT authentication"

# Step 1: Search for similar work
search_memory_facts(
    query="REST API JWT authentication implementation",
    group_ids=["python_development", "agent_executions"],
    max_facts=5
)

# Step 2: Review results
# Found: FastAPI + JWT implementation from 2 months ago
# Outcome: SUCCESS (94% test coverage, 120ms avg response)
# Lessons: "Use HTTP-only cookies", "Async context managers critical"

# Step 3: Apply insights
# - Use FastAPI (proven to work)
# - HTTP-only cookies for tokens
# - Async database connections
# - Start with pytest-asyncio fixtures

# Result: Faster implementation, fewer mistakes
```

### Stage 2: During Work

**Ask**: "Am I encountering a known problem?"

**Workflow**:
```
1. When error occurs, search error_resolutions
2. Check if similar issue solved before
3. Apply known solutions first
4. Try new approaches if no match
5. Document new resolution for future
```

**Example**:
```python
# Error encountered: "PostgreSQL connection pool exhausted"

# Step 1: Search for similar errors
search_memory