> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mengram.io/llms.txt
> Use this file to discover all available pages before exploring further.

# CrewAI

> Give your CrewAI agents persistent memory with procedural learning.

## Installation

```bash theme={null}
pip install mengram-ai[crewai]
```

## Quick start

```python theme={null}
from integrations.crewai import create_mengram_tools  # included in mengram-ai
from crewai import Agent, Task, Crew

# Create memory tools
tools = create_mengram_tools(api_key="om-your-key")

agent = Agent(
    role="Support Engineer",
    goal="Help users with technical issues using past context",
    tools=tools,
)

task = Task(
    description="Help the user debug their deployment issue",
    agent=agent,
)

crew = Crew(agents=[agent], tasks=[task])
result = crew.kickoff()
```

## Available tools

| Tool                        | Description                                                |
| --------------------------- | ---------------------------------------------------------- |
| `mengram_search`            | Search all 3 memory types (semantic, episodic, procedural) |
| `mengram_remember`          | Save information to memory (auto-extracts all 3 types)     |
| `mengram_profile`           | Get full user context via Cognitive Profile                |
| `mengram_save_workflow`     | Save a completed workflow as a procedure                   |
| `mengram_workflow_feedback` | Report success/failure of a workflow                       |

## Procedural learning

When a CrewAI agent completes a multi-step task, Mengram automatically saves it as a procedure. Next time a similar task comes up, the agent already knows the optimal path — with success/failure tracking.

```python theme={null}
# Agent workflow is automatically extracted as a procedure
# On next similar task, the agent retrieves the procedure
results = tools[0].run("how to deploy to Railway")
# Returns the learned procedure with steps
```