Flisk vs LangGraph and Mastra

LangGraph and Mastra build the agent. Flisk runs it next to its tools and database.

Task Flisk Local dev stack LangGraph Agent framework Mastra Agent framework
Start an agent project
flisk project agent-stack
flisk init agent-stack

Agent, tools, database and messaging

langgraph new my-app --template new-langgraph-project-python
npx create-mastra@latest my-app
Add an agent
flisk resource bot --type agent
No equivalent

Add the graph to langgraph.json by hand

No equivalent

Agents are defined in code

Add an MCP server
flisk resource tools --type mcp
No equivalent

mastra api mcp list only lists servers

Run locally with a dev UI
flisk up

Tilt UI shows the agent

langgraph dev

Prints a Studio URL

mastra dev

Serves Studio

List agent sessions
No equivalent

Threads are served by the API, not the CLI

Different job, closest command.
mastra api thread list

Lists memory threads on a running server

Build for production
No equivalent

Local only. Flisk never deploys

Run the built agent
langgraph up

Runs in Docker; builds the image unless --image is set

Write a Dockerfile
langgraph dockerfile Dockerfile
Connect a coding agent
Different job, closest command.
mastra init --mcp cursor

Init flag; editors only: cursor, windsurf, vscode