graphiti-episode-storage

# Graphiti Episode Storage

## Description

Patterns for storing episodes in Graphiti Memory including agent executions, error resolutions, workflow completions, and technical decisions. This skill provides proven JSON schemas and storage patterns for building institutional knowledge.

## When to Use

Automatically apply this skill when:
- Completing significant agent work
- Resolving non-trivial errors
- Finishing multi-step workflows
- Making important technical decisions
- Need to document execution for learning
- Creating audit trails

## Core Concept: Episodes

**Episodes** are discrete events stored in Graphiti Memory that get automatically processed into facts and entities:

- Agent executions and their outcomes
- Error resolutions with root causes
- Workflow completions with metrics
- Technical decision rationales

## Episode Storage Patterns

### Pattern 1: Agent Execution Episodes

Store each agent execution for learning and audit:

```python
mcp__graphiti-memory__add_memory(
    name=f"Agent Execution: {agent_name} - {task_summary}",
    episode_body=json.dumps({
        "agent": agent_name,
        "timestamp": datetime.now().isoformat(),
        "task_description": user_request,
        "context_provided": {
            "tech_stack": ["FastAPI", "PostgreSQL"],
            "constraints": ["Must support async", "90% test coverage"]
        },
        "approach_taken": [
            "Researched FastAPI async patterns",
            "Implemented connection pooling",
            "Added pytest-asyncio for testing"
        ],
        "outcome": "SUCCESS|PARTIAL|FAILED",
        "deliverables": [
            "API endpoints in /src/api/",
            "Tests in /tests/",
            "Documentation in /docs/"
        ],
        "lessons_learned": [
            "Async context managers critical for DB connections",
            "pytest-asyncio fixtures simplify testing"
        ],
        "time_spent_minutes": 45,
        "challenges_encountered": [
            "Initi